Mercurial > repos > goeckslab > image_learner

annotate plotly_plots.py @ 15:d17e3a1b8659 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
author goeckslab
date Fri, 28 Nov 2025 15:45:49 +0000
parents c5150cceab47
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
1 import json
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
2 from pathlib import Path
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
3 from typing import Dict, List, Optional
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
4
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
5 import numpy as np
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
6 import pandas as pd
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
7 import plotly.graph_objects as go
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
8 import plotly.io as pio
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
9 from constants import LABEL_COLUMN_NAME, SPLIT_COLUMN_NAME
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
10
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
11
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
12 def _style_fig(fig: go.Figure, font_size: int = 12) -> go.Figure:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
13 """Apply consistent styling across Plotly figures."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
14 fig.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
15 font=dict(size=font_size),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
16 plot_bgcolor="#ffffff",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
17 paper_bgcolor="#ffffff",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
18 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
19 fig.update_xaxes(gridcolor="#e8e8e8")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
20 fig.update_yaxes(gridcolor="#e8e8e8")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
21 return fig
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
22
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
23
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
24 def _labels_from_metadata_dict(meta_dict: dict) -> List[str]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
25 """Extract ordered label names from Ludwig train_set_metadata."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
26 if not isinstance(meta_dict, dict):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
27 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
28
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
29 for key in ("idx2str", "idx2label", "vocab"):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
30 seq = meta_dict.get(key)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
31 if isinstance(seq, list) and seq:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
32 return [str(v) for v in seq]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
33
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
34 str2idx = meta_dict.get("str2idx")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
35 if isinstance(str2idx, dict) and str2idx:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
36 int_indices = [v for v in str2idx.values() if isinstance(v, int)]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
37 if int_indices:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
38 max_idx = max(int_indices)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
39 ordered = [None] * (max_idx + 1)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
40 for name, idx in str2idx.items():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
41 if isinstance(idx, int) and 0 <= idx < len(ordered):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
42 ordered[idx] = name
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
43 return [str(v) for v in ordered if v is not None]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
44
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
45 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
46
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
47
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
48 def _resolve_confusion_labels(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
49 label_stats: dict,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
50 n_classes: int,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
51 metadata_csv_path: Optional[str],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
52 train_set_metadata_path: Optional[str],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
53 ) -> List[str]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
54 """Prefer original labels from metadata; fall back to stats if unavailable."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
55 if train_set_metadata_path:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
56 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
57 meta_path = Path(train_set_metadata_path)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
58 if meta_path.exists():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
59 with open(meta_path, "r") as f:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
60 meta_json = json.load(f)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
61 label_meta = meta_json.get(LABEL_COLUMN_NAME)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
62 if not isinstance(label_meta, dict):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
63 label_meta = next(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
64 (
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
65 v
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
66 for v in meta_json.values()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
67 if isinstance(v, dict)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
68 and any(k in v for k in ("idx2str", "str2idx", "idx2label", "vocab"))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
69 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
70 None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
71 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
72 labels_from_meta = _labels_from_metadata_dict(label_meta) if label_meta else []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
73 if labels_from_meta and len(labels_from_meta) >= n_classes:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
74 return [str(label) for label in labels_from_meta[:n_classes]]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
75 except Exception as exc:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
76 print(f"Warning: Unable to read labels from train_set_metadata: {exc}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
77
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
78 if metadata_csv_path:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
79 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
80 csv_path = Path(metadata_csv_path)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
81 if csv_path.exists():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
82 df_meta = pd.read_csv(csv_path)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
83 if LABEL_COLUMN_NAME in df_meta.columns:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
84 uniques = df_meta[LABEL_COLUMN_NAME].dropna().unique().tolist()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
85 if uniques and len(uniques) >= n_classes:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
86 return [str(u) for u in uniques[:n_classes]]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
87 except Exception as exc:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
88 print(f"Warning: Unable to read labels from metadata CSV: {exc}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
89
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
90 pcs = label_stats.get("per_class_stats", {})
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
91 if pcs:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
92 pcs_labels = [str(k) for k in pcs.keys()]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
93 if len(pcs_labels) >= n_classes:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
94 return pcs_labels[:n_classes]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
95
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
96 labels = label_stats.get("labels")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
97 if not labels:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
98 labels = [str(i) for i in range(n_classes)]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
99 if len(labels) < n_classes:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
100 labels = labels + [str(i) for i in range(len(labels), n_classes)]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
101 return [str(label) for label in labels[:n_classes]]
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
102
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
103
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
104 def build_classification_plots(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
105 test_stats_path: str,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
106 training_stats_path: Optional[str] = None,
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
107 metadata_csv_path: Optional[str] = None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
108 train_set_metadata_path: Optional[str] = None,
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
109 ) -> List[Dict[str, str]]:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
110 """
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
111 Read Ludwig’s test_statistics.json and build three interactive Plotly panels:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
112 - Confusion Matrix
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
113 - ROC-AUC
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
114 - Classification Report Heatmap
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
115
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
116 If metadata paths are provided, the confusion matrix axes will use the original
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
117 label values from the training metadata rather than integer-encoded labels.
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
118
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
119 Returns a list of dicts, each with:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
120 {
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
121 "title": <plot title>,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
122 "html": <HTML fragment for embedding>
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
123 }
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
124 """
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
125 # --- Load test stats ---
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
126 with open(test_stats_path, "r") as f:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
127 test_stats = json.load(f)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
128 label_stats = test_stats["label"]
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
129
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
130 # common sizing
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
131 cell = 40
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
132 n_classes = len(label_stats["confusion_matrix"])
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
133 side_px = max(cell * n_classes + 200, 600)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
134 common_cfg = {"displayModeBar": True, "scrollZoom": True}
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
135
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
136 plots: List[Dict[str, str]] = []
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
137
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
138 # 0) Confusion Matrix
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
139 cm = np.array(label_stats["confusion_matrix"], dtype=int)
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
140 labels = _resolve_confusion_labels(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
141 label_stats,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
142 n_classes,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
143 metadata_csv_path=metadata_csv_path,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
144 train_set_metadata_path=train_set_metadata_path,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
145 )
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
146 total = cm.sum()
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
147
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
148 fig_cm = go.Figure(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
149 go.Heatmap(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
150 z=cm,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
151 x=labels,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
152 y=labels,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
153 colorscale="Blues",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
154 showscale=True,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
155 colorbar=dict(title="Count"),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
156 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
157 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
158 fig_cm.update_traces(xgap=2, ygap=2)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
159 fig_cm.update_layout(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
160 title=dict(text="Confusion Matrix", x=0.5),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
161 xaxis_title="Predicted",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
162 yaxis_title="Observed",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
163 yaxis_autorange="reversed",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
164 width=side_px,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
165 height=side_px,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
166 margin=dict(t=100, l=80, r=80, b=80),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
167 )
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
168 _style_fig(fig_cm)
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
169
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
170 # annotate counts and percentages
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
171 mval = cm.max() if cm.size else 0
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
172 thresh = mval / 2
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
173 for i in range(cm.shape[0]):
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
174 for j in range(cm.shape[1]):
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
175 v = cm[i, j]
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
176 pct = (v / total * 100) if total > 0 else 0
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
177 color = "white" if v > thresh else "black"
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
178 fig_cm.add_annotation(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
179 x=labels[j],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
180 y=labels[i],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
181 text=f"<b>{v}</b>",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
182 showarrow=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
183 font=dict(color=color, size=14),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
184 xanchor="center",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
185 yanchor="bottom",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
186 yshift=2,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
187 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
188 fig_cm.add_annotation(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
189 x=labels[j],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
190 y=labels[i],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
191 text=f"{pct:.1f}%",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
192 showarrow=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
193 font=dict(color=color, size=13),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
194 xanchor="center",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
195 yanchor="top",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
196 yshift=-2,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
197 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
198
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
199 plots.append({
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
200 "title": "Confusion Matrix",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
201 "html": pio.to_html(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
202 fig_cm,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
203 full_html=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
204 include_plotlyjs="cdn",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
205 config=common_cfg
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
206 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
207 })
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
208
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
209 # 1) ROC Curve (from test_statistics)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
210 roc_plot = _build_static_roc_plot(label_stats, common_cfg, friendly_labels=labels)
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
211 if roc_plot:
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
212 plots.append(roc_plot)
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
213
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
214 # 2) Precision-Recall Curve (from test_statistics)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
215 pr_plot = _build_precision_recall_plot(label_stats, common_cfg)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
216 if pr_plot:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
217 plots.append(pr_plot)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
218
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
219 # 2) Classification Report Heatmap
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
220 pcs = label_stats.get("per_class_stats", {})
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
221 if pcs:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
222 classes = list(pcs.keys())
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
223 metrics = [
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
224 "precision",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
225 "recall",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
226 "f1_score",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
227 "accuracy",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
228 "matthews_correlation_coefficient",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
229 "specificity",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
230 ]
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
231 z, txt = [], []
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
232 for c in classes:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
233 row, trow = [], []
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
234 for m in metrics:
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
235 val = pcs[c].get(m, 0)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
236 row.append(val)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
237 trow.append(f"{val:.2f}")
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
238 z.append(row)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
239 txt.append(trow)
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
240
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
241 fig_cr = go.Figure(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
242 go.Heatmap(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
243 z=z,
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
244 x=[m.replace("_", " ") for m in metrics],
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
245 y=[str(c) for c in classes],
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
246 text=txt,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
247 texttemplate="%{text}",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
248 colorscale="Reds",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
249 showscale=True,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
250 colorbar=dict(title="Value"),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
251 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
252 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
253 fig_cr.update_layout(
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
254 title="Per-Class metrics",
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
255 xaxis_title="",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
256 yaxis_title="Class",
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
257 width=side_px,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
258 height=side_px,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
259 margin=dict(t=80, l=80, r=80, b=80),
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
260 )
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
261 _style_fig(fig_cr)
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
262 plots.append({
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
263 "title": "Per-Class metrics",
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
264 "html": pio.to_html(
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
265 fig_cr,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
266 full_html=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
267 include_plotlyjs=False,
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
268 config=common_cfg
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
269 )
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
270 })
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
271
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
272 # 3) Prediction Diagnostics (from predictions.csv)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
273 # Note: appended separately in generate_html_report, not returned here.
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
274
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
275 return plots
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
276
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
277
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
278 def build_train_validation_plots(train_stats_path: str) -> List[Dict[str, str]]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
279 """Generate Train/Validation learning curve plots from training_statistics.json."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
280 if not train_stats_path or not Path(train_stats_path).exists():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
281 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
282 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
283 with open(train_stats_path, "r") as f:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
284 train_stats = json.load(f)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
285 except Exception as exc:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
286 print(f"Warning: Unable to read training statistics: {exc}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
287 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
288
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
289 label_train = (train_stats.get("training") or {}).get("label", {})
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
290 label_val = (train_stats.get("validation") or {}).get("label", {})
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
291 if not label_train and not label_val:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
292 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
293 plots: List[Dict[str, str]] = []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
294 include_js = True # Load Plotly.js once for this group
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
295
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
296 def _get_series(stats: dict, metric: str) -> List[float]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
297 if metric not in stats:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
298 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
299 vals = stats.get(metric, [])
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
300 if isinstance(vals, list):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
301 return [float(v) for v in vals]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
302 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
303 return [float(vals)]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
304 except Exception:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
305 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
306
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
307 def _line_plot(metric_key: str, title: str, yaxis_title: str) -> Optional[Dict[str, str]]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
308 train_series = _get_series(label_train, metric_key)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
309 val_series = _get_series(label_val, metric_key)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
310 if not train_series and not val_series:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
311 return None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
312 epochs_train = list(range(1, len(train_series) + 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
313 epochs_val = list(range(1, len(val_series) + 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
314 fig = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
315 if train_series:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
316 fig.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
317 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
318 x=epochs_train,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
319 y=train_series,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
320 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
321 name="Train",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
322 line=dict(width=4),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
323 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
324 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
325 if val_series:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
326 fig.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
327 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
328 x=epochs_val,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
329 y=val_series,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
330 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
331 name="Validation",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
332 line=dict(width=4),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
333 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
334 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
335 fig.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
336 title=dict(text=title, x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
337 xaxis_title="Epoch",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
338 yaxis_title=yaxis_title,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
339 width=760,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
340 height=520,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
341 hovermode="x unified",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
342 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
343 _style_fig(fig)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
344 return {
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
345 "title": title,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
346 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
347 fig,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
348 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
349 include_plotlyjs="cdn" if include_js else False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
350 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
351 }
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
352
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
353 # Core learning curves
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
354 for key, title in [
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
355 ("roc_auc", "ROC-AUC across epochs"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
356 ("precision", "Precision across epochs"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
357 ("recall", "Recall/Sensitivity across epochs"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
358 ("specificity", "Specificity across epochs"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
359 ]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
360 plot = _line_plot(key, title, title.replace("Learning Curve", "").strip())
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
361 if plot:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
362 plots.append(plot)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
363 include_js = False
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
364
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
365 # Precision vs Recall evolution (validation)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
366 val_prec = _get_series(label_val, "precision")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
367 val_rec = _get_series(label_val, "recall")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
368 if val_prec and val_rec:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
369 epochs = list(range(1, min(len(val_prec), len(val_rec)) + 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
370 fig_pr = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
371 fig_pr.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
372 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
373 x=epochs,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
374 y=val_prec[: len(epochs)],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
375 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
376 name="Precision",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
377 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
378 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
379 fig_pr.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
380 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
381 x=epochs,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
382 y=val_rec[: len(epochs)],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
383 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
384 name="Recall",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
385 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
386 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
387 fig_pr.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
388 title=dict(text="Validation Precision and Recall by Epoch", x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
389 xaxis_title="Epoch",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
390 yaxis_title="Value",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
391 width=760,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
392 height=520,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
393 hovermode="x unified",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
394 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
395 _style_fig(fig_pr)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
396 plots.append({
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
397 "title": "Precision vs Recall Evolution",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
398 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
399 fig_pr,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
400 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
401 include_plotlyjs="cdn" if include_js else False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
402 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
403 })
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
404 include_js = False
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
405
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
406 # F1-score derived
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
407 def _compute_f1(p: List[float], r: List[float]) -> List[float]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
408 f1_vals = []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
409 for prec, rec in zip(p, r):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
410 if (prec + rec) == 0:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
411 f1_vals.append(0.0)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
412 else:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
413 f1_vals.append(2 * prec * rec / (prec + rec))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
414 return f1_vals
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
415
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
416 f1_train = _compute_f1(_get_series(label_train, "precision"), _get_series(label_train, "recall"))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
417 f1_val = _compute_f1(val_prec, val_rec)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
418 if f1_train or f1_val:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
419 fig = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
420 if f1_train:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
421 fig.add_trace(go.Scatter(x=list(range(1, len(f1_train) + 1)), y=f1_train, mode="lines+markers", name="Train", line=dict(width=4)))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
422 if f1_val:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
423 fig.add_trace(go.Scatter(x=list(range(1, len(f1_val) + 1)), y=f1_val, mode="lines+markers", name="Validation", line=dict(width=4)))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
424 fig.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
425 title=dict(text="F1-Score across epochs (derived)", x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
426 xaxis_title="Epoch",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
427 yaxis_title="F1-Score",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
428 width=760,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
429 height=520,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
430 hovermode="x unified",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
431 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
432 _style_fig(fig)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
433 plots.append({
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
434 "title": "F1-Score across epochs (derived)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
435 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
436 fig,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
437 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
438 include_plotlyjs="cdn" if include_js else False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
439 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
440 })
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
441 include_js = False
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
442
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
443 # Overfitting Gap: Train vs Val ROC-AUC (gap)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
444 roc_train = _get_series(label_train, "roc_auc")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
445 roc_val = _get_series(label_val, "roc_auc")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
446 if roc_train and roc_val:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
447 epochs_gap = list(range(1, min(len(roc_train), len(roc_val)) + 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
448 gaps = [t - v for t, v in zip(roc_train[:len(epochs_gap)], roc_val[:len(epochs_gap)])]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
449 fig_gap = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
450 fig_gap.add_trace(go.Scatter(x=epochs_gap, y=gaps, mode="lines+markers", name="Train - Val ROC-AUC", line=dict(width=4)))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
451 fig_gap.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
452 title=dict(text="Overfitting gap: ROC-AUC across epochs", x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
453 xaxis_title="Epoch",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
454 yaxis_title="Gap",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
455 width=760,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
456 height=520,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
457 hovermode="x unified",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
458 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
459 _style_fig(fig_gap)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
460 plots.append({
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
461 "title": "Overfitting gap: ROC-AUC across epochs",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
462 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
463 fig_gap,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
464 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
465 include_plotlyjs="cdn" if include_js else False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
466 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
467 })
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
468 include_js = False
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
469
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
470 # Best Epoch Dashboard (based on max val ROC-AUC)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
471 if roc_val:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
472 best_idx = int(np.argmax(roc_val))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
473 best_epoch = best_idx + 1
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
474 spec_val = _get_series(label_val, "specificity")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
475 metrics_at_best = {
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
476 "ROC-AUC": roc_val[best_idx] if best_idx < len(roc_val) else None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
477 "Precision": val_prec[best_idx] if best_idx < len(val_prec) else None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
478 "Recall": val_rec[best_idx] if best_idx < len(val_rec) else None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
479 "Specificity": spec_val[best_idx] if best_idx < len(spec_val) else None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
480 "F1-Score": f1_val[best_idx] if best_idx < len(f1_val) else None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
481 }
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
482 fig_best = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
483 for name, value in metrics_at_best.items():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
484 if value is not None:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
485 fig_best.add_trace(go.Bar(name=name, x=[name], y=[value]))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
486 fig_best.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
487 title=dict(text=f"Best Epoch Dashboard (Val ROC-AUC @ epoch {best_epoch})", x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
488 xaxis_title="Metric",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
489 yaxis_title="Value",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
490 width=760,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
491 height=520,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
492 showlegend=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
493 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
494 _style_fig(fig_best)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
495 plots.append({
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
496 "title": "Best Validation Epoch Snapshot (Metrics)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
497 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
498 fig_best,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
499 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
500 include_plotlyjs="cdn" if include_js else False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
501 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
502 })
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
503 include_js = False
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
504
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents:
diff changeset
505 return plots
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
506
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
507
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
508 def _get_regression_series(split_stats: dict, metric: str) -> List[float]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
509 if metric not in split_stats:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
510 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
511 vals = split_stats.get(metric, [])
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
512 if isinstance(vals, list):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
513 return [float(v) for v in vals]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
514 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
515 return [float(vals)]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
516 except Exception:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
517 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
518
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
519
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
520 def _regression_line_plot(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
521 train_split: dict,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
522 val_split: dict,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
523 metric_key: str,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
524 title: str,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
525 yaxis_title: str,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
526 include_js: bool,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
527 ) -> Optional[Dict[str, str]]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
528 train_series = _get_regression_series(train_split, metric_key)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
529 val_series = _get_regression_series(val_split, metric_key)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
530 if not train_series and not val_series:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
531 return None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
532 epochs_train = list(range(1, len(train_series) + 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
533 epochs_val = list(range(1, len(val_series) + 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
534 fig = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
535 if train_series:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
536 fig.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
537 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
538 x=epochs_train,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
539 y=train_series,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
540 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
541 name="Train",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
542 line=dict(width=4),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
543 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
544 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
545 if val_series:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
546 fig.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
547 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
548 x=epochs_val,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
549 y=val_series,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
550 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
551 name="Validation",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
552 line=dict(width=4),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
553 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
554 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
555 fig.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
556 title=dict(text=title, x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
557 xaxis_title="Epoch",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
558 yaxis_title=yaxis_title,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
559 width=760,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
560 height=520,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
561 hovermode="x unified",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
562 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
563 _style_fig(fig)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
564 return {
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
565 "title": title,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
566 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
567 fig,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
568 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
569 include_plotlyjs="cdn" if include_js else False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
570 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
571 }
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
572
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
573
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
574 def build_regression_train_val_plots(train_stats_path: str) -> List[Dict[str, str]]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
575 """Generate regression Train/Validation learning curve plots from training_statistics.json."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
576 if not train_stats_path or not Path(train_stats_path).exists():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
577 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
578 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
579 with open(train_stats_path, "r") as f:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
580 train_stats = json.load(f)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
581 except Exception as exc:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
582 print(f"Warning: Unable to read training statistics: {exc}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
583 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
584
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
585 label_train = (train_stats.get("training") or {}).get("label", {})
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
586 label_val = (train_stats.get("validation") or {}).get("label", {})
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
587 if not label_train and not label_val:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
588 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
589
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
590 plots: List[Dict[str, str]] = []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
591 include_js = True
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
592 for metric_key, title, ytitle in [
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
593 ("mean_absolute_error", "Mean Absolute Error across epochs", "MAE"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
594 ("root_mean_squared_error", "Root Mean Squared Error across epochs", "RMSE"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
595 ("mean_absolute_percentage_error", "Mean Absolute Percentage Error across epochs", "MAPE"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
596 ("r2", "R² across epochs", "R²"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
597 ("loss", "Loss across epochs", "Loss"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
598 ]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
599 plot = _regression_line_plot(label_train, label_val, metric_key, title, ytitle, include_js)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
600 if plot:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
601 plots.append(plot)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
602 include_js = False
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
603 return plots
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
604
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
605
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
606 def build_regression_test_plots(train_stats_path: str) -> List[Dict[str, str]]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
607 """Generate regression Test learning curves from training_statistics.json."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
608 if not train_stats_path or not Path(train_stats_path).exists():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
609 return []
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
610 try:
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
611 with open(train_stats_path, "r") as f:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
612 train_stats = json.load(f)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
613 except Exception as exc:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
614 print(f"Warning: Unable to read training statistics: {exc}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
615 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
616
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
617 label_test = (train_stats.get("test") or {}).get("label", {})
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
618 if not label_test:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
619 return []
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
620
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
621 plots: List[Dict[str, str]] = []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
622 include_js = True
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
623 metrics = [
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
624 ("mean_absolute_error", "Mean Absolute Error Across Epochs", "MAE"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
625 ("root_mean_squared_error", "Root Mean Squared Error Across Epochs", "RMSE"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
626 ("mean_absolute_percentage_error", "Mean Absolute Percentage Error Across Epochs", "MAPE"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
627 ("r2", "R² Across Epochs", "R²"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
628 ("loss", "Loss Across Epochs", "Loss"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
629 ]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
630 epochs = None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
631 for metric_key, title, ytitle in metrics:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
632 series = _get_regression_series(label_test, metric_key)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
633 if not series:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
634 continue
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
635 if epochs is None:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
636 epochs = list(range(1, len(series) + 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
637 fig = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
638 fig.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
639 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
640 x=epochs,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
641 y=series[: len(epochs)],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
642 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
643 name="Test",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
644 line=dict(width=4),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
645 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
646 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
647 fig.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
648 title=dict(text=title, x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
649 xaxis_title="Epoch",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
650 yaxis_title=ytitle,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
651 width=760,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
652 height=520,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
653 hovermode="x unified",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
654 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
655 _style_fig(fig)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
656 plots.append({
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
657 "title": title,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
658 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
659 fig,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
660 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
661 include_plotlyjs="cdn" if include_js else False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
662 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
663 })
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
664 include_js = False
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
665 return plots
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
666
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
667
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
668 def _build_static_roc_plot(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
669 label_stats: dict, config: dict, friendly_labels: Optional[List[str]] = None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
670 ) -> Optional[Dict[str, str]]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
671 """Build ROC curve directly from test_statistics.json (single curve)."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
672 roc_data = label_stats.get("roc_curve")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
673 if not isinstance(roc_data, dict):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
674 return None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
675
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
676 fpr = roc_data.get("false_positive_rate")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
677 tpr = roc_data.get("true_positive_rate")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
678 if not fpr or not tpr or len(fpr) != len(tpr):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
679 return None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
680
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
681 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
682 fig = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
683 fig.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
684 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
685 x=fpr,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
686 y=tpr,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
687 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
688 name="ROC Curve",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
689 line=dict(color="#1f77b4", width=4),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
690 hovertemplate="FPR: %{x:.3f}<br>TPR: %{y:.3f}<extra></extra>",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
691 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
692 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
693 fig.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
694 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
695 x=[0, 1],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
696 y=[0, 1],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
697 mode="lines",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
698 name="Random Classifier",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
699 line=dict(color="gray", width=2, dash="dash"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
700 hovertemplate="Random Classifier<extra></extra>",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
701 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
702 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
703
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
704 auc_val = label_stats.get("roc_auc") or label_stats.get("roc_auc_macro") or label_stats.get("roc_auc_micro")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
705 auc_txt = f" (AUC = {auc_val:.3f})" if isinstance(auc_val, (int, float)) else ""
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
706
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
707 # Determine which label is treated as positive for the curve
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
708 label_list: List = []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
709 pcs = label_stats.get("per_class_stats", {})
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
710 if pcs:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
711 label_list = list(pcs.keys())
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
712 if not label_list:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
713 labels_from_stats = label_stats.get("labels")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
714 if isinstance(labels_from_stats, list):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
715 label_list = labels_from_stats
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
716
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
717 # Try to resolve index of the positive label explicitly provided by Ludwig
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
718 pos_label_raw = (
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
719 roc_data.get("positive_label")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
720 or roc_data.get("positive_class")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
721 or label_stats.get("positive_label")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
722 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
723 pos_label_idx = None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
724 if pos_label_raw is not None and isinstance(label_list, list):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
725 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
726 pos_label_idx = label_list.index(pos_label_raw)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
727 except ValueError:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
728 pos_label_idx = None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
729
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
730 # Fallback: use the second label if available, otherwise the first
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
731 if pos_label_idx is None:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
732 if isinstance(label_list, list) and len(label_list) >= 2:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
733 pos_label_idx = 1
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
734 elif isinstance(label_list, list) and label_list:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
735 pos_label_idx = 0
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
736
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
737 if pos_label_raw is None and isinstance(label_list, list) and pos_label_idx is not None:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
738 pos_label_raw = label_list[pos_label_idx]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
739
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
740 # Map to friendly label if we have one from metadata/CSV
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
741 pos_label_display = pos_label_raw
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
742 if (
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
743 friendly_labels
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
744 and isinstance(pos_label_idx, int)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
745 and 0 <= pos_label_idx < len(friendly_labels)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
746 ):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
747 pos_label_display = friendly_labels[pos_label_idx]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
748
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
749 pos_label_txt = (
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
750 f"Positive class: {pos_label_display}"
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
751 if pos_label_display is not None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
752 else "Positive class: (not available)"
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
753 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
754
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
755 title_label = f"ROC Curve{auc_txt}"
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
756 if pos_label_display is not None:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
757 title_label = f"ROC Curve (Positive Class: {pos_label_display}){auc_txt}"
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
758
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
759 fig.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
760 title=dict(text=title_label, x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
761 xaxis_title="False Positive Rate",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
762 yaxis_title="True Positive Rate",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
763 width=700,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
764 height=600,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
765 margin=dict(t=80, l=80, r=80, b=110),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
766 hovermode="closest",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
767 legend=dict(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
768 x=0.6,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
769 y=0.1,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
770 bgcolor="rgba(255,255,255,0.9)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
771 bordercolor="rgba(0,0,0,0.2)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
772 borderwidth=1,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
773 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
774 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
775 _style_fig(fig)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
776 fig.update_xaxes(range=[0, 1.0])
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
777 fig.update_yaxes(range=[0, 1.05])
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
778
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
779 fig.add_annotation(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
780 x=0.5,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
781 y=-0.15,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
782 xref="paper",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
783 yref="paper",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
784 showarrow=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
785 text=f"<span style='font-size:12px;color:#555;'>{pos_label_txt}</span>",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
786 xanchor="center",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
787 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
788
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
789 return {
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
790 "title": "ROC Curve",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
791 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
792 fig,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
793 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
794 include_plotlyjs=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
795 config=config,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
796 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
797 }
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
798 except Exception as e:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
799 print(f"Error building ROC plot: {e}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
800 return None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
801
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
802
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
803 def _build_precision_recall_plot(label_stats: dict, config: dict) -> Optional[Dict[str, str]]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
804 """Build Precision-Recall curve directly from test_statistics.json."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
805 pr_data = label_stats.get("precision_recall_curve")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
806 if not isinstance(pr_data, dict):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
807 return None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
808
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
809 precisions = pr_data.get("precisions")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
810 recalls = pr_data.get("recalls")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
811 if not precisions or not recalls or len(precisions) != len(recalls):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
812 return None
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
813
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
814 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
815 fig = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
816 fig.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
817 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
818 x=recalls,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
819 y=precisions,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
820 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
821 name="Precision-Recall",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
822 line=dict(color="#d62728", width=4),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
823 hovertemplate="Recall: %{x:.3f}<br>Precision: %{y:.3f}<extra></extra>",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
824 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
825 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
826
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
827 ap_val = (
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
828 label_stats.get("average_precision_macro")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
829 or label_stats.get("average_precision_micro")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
830 or label_stats.get("average_precision_samples")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
831 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
832 ap_txt = f" (AP = {ap_val:.3f})" if isinstance(ap_val, (int, float)) else ""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
833
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
834 fig.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
835 title=dict(text=f"Precision-Recall Curve{ap_txt}", x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
836 xaxis_title="Recall",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
837 yaxis_title="Precision",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
838 width=700,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
839 height=600,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
840 margin=dict(t=80, l=80, r=80, b=80),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
841 hovermode="closest",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
842 legend=dict(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
843 x=0.6,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
844 y=0.1,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
845 bgcolor="rgba(255,255,255,0.9)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
846 bordercolor="rgba(0,0,0,0.2)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
847 borderwidth=1,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
848 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
849 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
850 _style_fig(fig)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
851 fig.update_xaxes(range=[0, 1.0])
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
852 fig.update_yaxes(range=[0, 1.05])
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
853
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
854 return {
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
855 "title": "Precision-Recall Curve",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
856 "html": pio.to_html(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
857 fig,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
858 full_html=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
859 include_plotlyjs=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
860 config=config,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
861 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
862 }
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
863 except Exception as e:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
864 print(f"Error building Precision-Recall plot: {e}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
865 return None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
866
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
867
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
868 def build_prediction_diagnostics(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
869 predictions_path: str,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
870 label_data_path: Optional[str] = None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
871 split_value: int = 2,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
872 threshold: Optional[float] = None,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
873 ) -> List[Dict[str, str]]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
874 """Generate diagnostic plots from predictions.csv for classification tasks."""
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
875 preds_file = Path(predictions_path)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
876 if not preds_file.exists():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
877 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
878
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
879 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
880 df_pred = pd.read_csv(predictions_path)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
881 except Exception as exc:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
882 print(f"Warning: Unable to read predictions CSV: {exc}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
883 return []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
884
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
885 plots: List[Dict[str, str]] = []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
886
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
887 # Identify probability columns
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
888 prob_cols = [
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
889 c for c in df_pred.columns
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
890 if c.startswith("label_probabilities_") and c != "label_probabilities"
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
891 ]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
892 prob_cols_sorted = sorted(prob_cols)
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
893
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
894 def _select_positive_prob():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
895 if not prob_cols_sorted:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
896 return None, None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
897 # Prefer a column indicating positive/event/true/1
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
898 preferred_keys = ("event", "true", "positive", "pos", "1")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
899 for col in prob_cols_sorted:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
900 suffix = col.replace("label_probabilities_", "").lower()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
901 if any(k in suffix for k in preferred_keys):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
902 return col, suffix
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
903 if len(prob_cols_sorted) == 2:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
904 col = prob_cols_sorted[1]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
905 return col, col.replace("label_probabilities_", "")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
906 col = prob_cols_sorted[0]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
907 return col, col.replace("label_probabilities_", "")
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
908
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
909 pos_prob_col, pos_label_hint = _select_positive_prob()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
910 pos_prob_series = df_pred[pos_prob_col] if pos_prob_col and pos_prob_col in df_pred else None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
911
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
912 # Confidence series: prefer label_probability, otherwise positive prob, otherwise max prob
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
913 confidence_series = None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
914 if "label_probability" in df_pred.columns:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
915 confidence_series = df_pred["label_probability"]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
916 elif pos_prob_series is not None:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
917 confidence_series = pos_prob_series
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
918 elif prob_cols_sorted:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
919 confidence_series = df_pred[prob_cols_sorted].max(axis=1)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
920
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
921 # True labels
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
922 def _extract_labels():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
923 candidates = [
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
924 LABEL_COLUMN_NAME,
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
925 f"{LABEL_COLUMN_NAME}_ground_truth",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
926 f"{LABEL_COLUMN_NAME}__ground_truth",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
927 f"{LABEL_COLUMN_NAME}_target",
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
928 f"{LABEL_COLUMN_NAME}__target",
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
929 "label",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
930 "label_true",
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
931 ]
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
932 candidates.extend(
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
933 [
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
934 col
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
935 for col in df_pred.columns
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
936 if (col.startswith(f"{LABEL_COLUMN_NAME}_") or col.startswith(f"{LABEL_COLUMN_NAME}__"))
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
937 and "probabilities" not in col
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
938 and "predictions" not in col
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
939 ]
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
940 )
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
941 for col in candidates:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
942 if col in df_pred.columns and col not in prob_cols_sorted:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
943 return df_pred[col]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
944 if label_data_path and Path(label_data_path).exists():
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
945 try:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
946 df_all = pd.read_csv(label_data_path)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
947 if SPLIT_COLUMN_NAME in df_all.columns:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
948 df_all = df_all[df_all[SPLIT_COLUMN_NAME] == split_value].reset_index(drop=True)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
949 if LABEL_COLUMN_NAME in df_all.columns:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
950 return df_all[LABEL_COLUMN_NAME].reset_index(drop=True)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
951 except Exception as exc:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
952 print(f"Warning: Unable to load labels from dataset: {exc}")
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
953 return None
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
954
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
955 labels_series = _extract_labels()
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
956
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
957 # Plot 1: Confidence Histogram
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
958 if confidence_series is not None:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
959 fig_conf = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
960 fig_conf.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
961 go.Histogram(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
962 x=confidence_series,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
963 nbinsx=20,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
964 marker=dict(color="#1f77b4", line=dict(color="#ffffff", width=1)),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
965 opacity=0.8,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
966 histnorm="percent",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
967 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
968 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
969 fig_conf.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
970 title=dict(text="Prediction Confidence Distribution", x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
971 xaxis_title="Predicted probability (confidence)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
972 yaxis_title="Percentage (%)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
973 bargap=0.05,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
974 width=700,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
975 height=500,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
976 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
977 _style_fig(fig_conf)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
978 plots.append({
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
979 "title": "Prediction Confidence Distribution",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
980 "html": pio.to_html(fig_conf, full_html=False, include_plotlyjs=False),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
981 })
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
982
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
983 # The remaining plots require true labels and a positive-class probability
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
984 if labels_series is None or pos_prob_series is None:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
985 return plots
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
986
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
987 # Align lengths
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
988 min_len = min(len(labels_series), len(pos_prob_series))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
989 if min_len == 0:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
990 return plots
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
991 y_true_raw = labels_series.iloc[:min_len]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
992 y_score = np.array(pos_prob_series.iloc[:min_len], dtype=float)
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
993
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
994 # Determine positive label
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
995 unique_labels = pd.unique(y_true_raw)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
996 unique_labels_list = list(unique_labels)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
997 positive_label = None
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
998 if pos_label_hint and str(pos_label_hint) in [str(u) for u in unique_labels_list]:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
999 positive_label = pos_label_hint
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1000 elif len(unique_labels_list) == 2:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1001 positive_label = unique_labels_list[1]
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1002 else:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1003 positive_label = unique_labels_list[0]
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
1004
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1005 y_true = (y_true_raw == positive_label).astype(int).values
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
1006
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1007 # Plot 2: Calibration Curve
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1008 bins = np.linspace(0.0, 1.0, 11)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1009 bin_ids = np.digitize(y_score, bins, right=True)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1010 bin_centers = []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1011 frac_positives = []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1012 for b in range(1, len(bins)):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1013 mask = bin_ids == b
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1014 if not np.any(mask):
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1015 continue
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1016 bin_centers.append(y_score[mask].mean())
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1017 frac_positives.append(y_true[mask].mean())
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1018 if bin_centers and frac_positives:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1019 fig_cal = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1020 fig_cal.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1021 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1022 x=bin_centers,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1023 y=frac_positives,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1024 mode="lines+markers",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1025 name="Calibration",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1026 line=dict(color="#2ca02c", width=4),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1027 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1028 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1029 fig_cal.add_trace(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1030 go.Scatter(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1031 x=[0, 1],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1032 y=[0, 1],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1033 mode="lines",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1034 name="Perfect Calibration",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1035 line=dict(color="gray", width=2, dash="dash"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1036 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1037 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1038 fig_cal.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1039 title=dict(text="Calibration Curve", x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1040 xaxis_title="Predicted probability",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1041 yaxis_title="Observed frequency",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1042 width=700,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1043 height=500,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1044 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1045 _style_fig(fig_cal)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1046 plots.append({
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1047 "title": "Calibration Curve (Predicted Probability vs Observed Frequency)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1048 "html": pio.to_html(fig_cal, full_html=False, include_plotlyjs=False),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1049 })
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
1050
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1051 # Plot 3: Threshold vs Metrics
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1052 thresholds = np.linspace(0.0, 1.0, 21)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1053 accs, f1s, sens, specs = [], [], [], []
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1054 for t in thresholds:
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1055 y_pred = (y_score >= t).astype(int)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1056 tp = np.sum((y_true == 1) & (y_pred == 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1057 tn = np.sum((y_true == 0) & (y_pred == 0))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1058 fp = np.sum((y_true == 0) & (y_pred == 1))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1059 fn = np.sum((y_true == 1) & (y_pred == 0))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1060 acc = (tp + tn) / max(len(y_true), 1)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1061 prec = tp / max(tp + fp, 1e-9)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1062 rec = tp / max(tp + fn, 1e-9)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1063 f1 = 0 if (prec + rec) == 0 else 2 * prec * rec / (prec + rec)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1064 sensitivity = rec
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1065 specificity = tn / max(tn + fp, 1e-9)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1066 accs.append(acc)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1067 f1s.append(f1)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1068 sens.append(sensitivity)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1069 specs.append(specificity)
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
1070
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1071 fig_thresh = go.Figure()
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1072 fig_thresh.add_trace(go.Scatter(x=thresholds, y=accs, mode="lines", name="Accuracy", line=dict(width=4)))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1073 fig_thresh.add_trace(go.Scatter(x=thresholds, y=f1s, mode="lines", name="F1", line=dict(width=4)))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1074 fig_thresh.add_trace(go.Scatter(x=thresholds, y=sens, mode="lines", name="Sensitivity", line=dict(width=4)))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1075 fig_thresh.add_trace(go.Scatter(x=thresholds, y=specs, mode="lines", name="Specificity", line=dict(width=4)))
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1076 fig_thresh.update_layout(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1077 title=dict(text="Threshold Sweep: Accuracy, F1, Sensitivity, Specificity", x=0.5),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1078 xaxis_title="Decision threshold",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1079 yaxis_title="Metric value",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1080 width=700,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1081 height=500,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1082 legend=dict(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1083 x=0.7,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1084 y=0.2,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1085 bgcolor="rgba(255,255,255,0.9)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1086 bordercolor="rgba(0,0,0,0.2)",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1087 borderwidth=1,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1088 ),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1089 shapes=[
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1090 dict(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1091 type="line",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1092 x0=threshold,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1093 x1=threshold,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1094 y0=0,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1095 y1=1,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1096 xref="x",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1097 yref="paper",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1098 line=dict(color="#d62728", width=2, dash="dash"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1099 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1100 ] if isinstance(threshold, (int, float)) else [],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1101 annotations=[
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1102 dict(
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1103 x=threshold,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1104 y=1.02,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1105 xref="x",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1106 yref="paper",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1107 showarrow=False,
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1108 text=f"Threshold = {threshold:.2f}",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1109 font=dict(size=11, color="#d62728"),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1110 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1111 ] if isinstance(threshold, (int, float)) else [],
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1112 )
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1113 _style_fig(fig_thresh)
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1114 plots.append({
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1115 "title": "Threshold Sweep: Accuracy, F1, Sensitivity, Specificity",
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1116 "html": pio.to_html(fig_thresh, full_html=False, include_plotlyjs=False),
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1117 })
11
c5150cceab47 planemo upload for repository https://github.com/goeckslab/gleam.git commit 0fe927b618cd4dfc87af7baaa827034cc6813225
goeckslab
parents: 8
diff changeset
1118
15
d17e3a1b8659 planemo upload for repository https://github.com/goeckslab/gleam.git commit bc50fef8acb44aca15d0a1746e6c0c967da5bb17
goeckslab
parents: 11
diff changeset
1119 return plots