Mercurial > repos > goeckslab > image_learner

annotate utils.py @ 13:1a9c42974a5a 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 9f96da4ea7ab3b572af86698ff51b870125cd674
author goeckslab
date Fri, 21 Nov 2025 17:35:00 +0000
parents bcfa2e234a80
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
12
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
1 import logging
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
2 from pathlib import Path
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
3
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
4 import pandas as pd
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
5
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
6 logger = logging.getLogger("ImageLearner")
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
7
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
8
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
9 def load_metadata_table(file_path: Path) -> pd.DataFrame:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
10 """Load image metadata allowing either CSV or TSV delimiters."""
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
11 logger.info("Loading metadata table from %s", file_path)
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
12 return pd.read_csv(file_path, sep=None, engine="python")
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
13
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
14
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
15 def detect_output_type(test_stats):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
16 """Detects if the output type is 'binary' or 'category' based on test statistics."""
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
17 label_stats = test_stats.get("label", {})
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
18 if "mean_squared_error" in label_stats:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
19 return "regression"
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
20 per_class = label_stats.get("per_class_stats", {})
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
21 if len(per_class) == 2:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
22 return "binary"
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
23 return "category"
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
24
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
25
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
26 def aug_parse(aug_string: str):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
27 """
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
28 Parse comma-separated augmentation keys into Ludwig augmentation dicts.
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
29 Raises ValueError on unknown key.
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
30 """
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
31 mapping = {
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
32 "random_horizontal_flip": {"type": "random_horizontal_flip"},
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
33 "random_vertical_flip": {"type": "random_vertical_flip"},
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
34 "random_rotate": {"type": "random_rotate", "degree": 10},
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
35 "random_blur": {"type": "random_blur", "kernel_size": 3},
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
36 "random_brightness": {"type": "random_brightness", "min": 0.5, "max": 2.0},
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
37 "random_contrast": {"type": "random_contrast", "min": 0.5, "max": 2.0},
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
38 }
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
39 aug_list = []
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
40 for tok in aug_string.split(","):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
41 key = tok.strip()
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
42 if not key:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
43 continue
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
44 if key not in mapping:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
45 valid = ", ".join(mapping.keys())
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
46 raise ValueError(f"Unknown augmentation '{key}'. Valid choices: {valid}")
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
47 aug_list.append(mapping[key])
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
48 return aug_list
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
49
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
50
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
51 def argument_checker(args, parser):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
52 if not 0.0 <= args.validation_size <= 1.0:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
53 parser.error("validation-size must be between 0.0 and 1.0")
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
54 if not args.csv_file.is_file():
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
55 parser.error(f"Metada file not found: {args.csv_file}")
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
56 if not (args.image_zip.is_file() or args.image_zip.is_dir()):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
57 parser.error(f"ZIP or directory not found: {args.image_zip}")
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
58 if args.augmentation is not None:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
59 try:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
60 augmentation_setup = aug_parse(args.augmentation)
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
61 setattr(args, "augmentation", augmentation_setup)
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
62 except ValueError as e:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
63 parser.error(str(e))
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
64
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
65
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
66 def parse_learning_rate(s):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
67 try:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
68 return float(s)
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
69 except (TypeError, ValueError):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
70 return None
0
54b871dfc51e planemo upload for repository https://github.com/goeckslab/gleam.git commit b7411ff35b6228ccdfd36cd4ebd946c03ac7f7e9
goeckslab
parents:
diff changeset
71
54b871dfc51e planemo upload for repository https://github.com/goeckslab/gleam.git commit b7411ff35b6228ccdfd36cd4ebd946c03ac7f7e9
goeckslab
parents:
diff changeset
72
12
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
73 def extract_metrics_from_json(
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
74 train_stats: dict,
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
75 test_stats: dict,
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
76 output_type: str,
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
77 ) -> dict:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
78 """Extracts relevant metrics from training and test statistics based on the output type."""
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
79 metrics = {"training": {}, "validation": {}, "test": {}}
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents: 2
diff changeset
80
12
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
81 def get_last_value(stats, key):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
82 val = stats.get(key)
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
83 if isinstance(val, list) and val:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
84 return val[-1]
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
85 elif isinstance(val, (int, float)):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
86 return val
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
87 return None
8
85e6f4b2ad18 planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
goeckslab
parents: 2
diff changeset
88
12
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
89 for split in ["training", "validation"]:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
90 split_stats = train_stats.get(split, {})
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
91 if not split_stats:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
92 logger.warning("No statistics found for %s split", split)
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
93 continue
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
94 label_stats = split_stats.get("label", {})
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
95 if not label_stats:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
96 logger.warning("No label statistics found for %s split", split)
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
97 continue
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
98 if output_type == "binary":
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
99 metrics[split] = {
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
100 "accuracy": get_last_value(label_stats, "accuracy"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
101 "loss": get_last_value(label_stats, "loss"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
102 "precision": get_last_value(label_stats, "precision"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
103 "recall": get_last_value(label_stats, "recall"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
104 "specificity": get_last_value(label_stats, "specificity"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
105 "roc_auc": get_last_value(label_stats, "roc_auc"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
106 }
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
107 elif output_type == "regression":
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
108 metrics[split] = {
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
109 "loss": get_last_value(label_stats, "loss"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
110 "mean_absolute_error": get_last_value(
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
111 label_stats, "mean_absolute_error"
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
112 ),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
113 "mean_absolute_percentage_error": get_last_value(
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
114 label_stats, "mean_absolute_percentage_error"
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
115 ),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
116 "mean_squared_error": get_last_value(label_stats, "mean_squared_error"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
117 "root_mean_squared_error": get_last_value(
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
118 label_stats, "root_mean_squared_error"
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
119 ),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
120 "root_mean_squared_percentage_error": get_last_value(
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
121 label_stats, "root_mean_squared_percentage_error"
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
122 ),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
123 "r2": get_last_value(label_stats, "r2"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
124 }
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
125 else:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
126 metrics[split] = {
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
127 "accuracy": get_last_value(label_stats, "accuracy"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
128 "accuracy_micro": get_last_value(label_stats, "accuracy_micro"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
129 "loss": get_last_value(label_stats, "loss"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
130 "roc_auc": get_last_value(label_stats, "roc_auc"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
131 "hits_at_k": get_last_value(label_stats, "hits_at_k"),
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
132 }
2
186424a7eca7 planemo upload for repository https://github.com/goeckslab/gleam.git commit 91fa4aba245520fc0680088a07cead66bcfd4ed2
goeckslab
parents: 1
diff changeset
133
12
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
134 # Test metrics: dynamic extraction according to exclusions
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
135 test_label_stats = test_stats.get("label", {})
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
136 if not test_label_stats:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
137 logger.warning("No label statistics found for test split")
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
138 else:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
139 combined_stats = test_stats.get("combined", {})
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
140 overall_stats = test_label_stats.get("overall_stats", {})
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
141
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
142 # Define exclusions
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
143 if output_type == "binary":
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
144 exclude = {"per_class_stats", "precision_recall_curve", "roc_curve"}
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
145 else:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
146 exclude = {"per_class_stats", "confusion_matrix"}
10
b0d893d04d4c planemo upload for repository https://github.com/goeckslab/gleam.git commit 1594d503179f28987720594eb49b48a15486f073
goeckslab
parents: 9
diff changeset
147
12
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
148 # 1. Get all scalar test_label_stats not excluded
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
149 test_metrics = {}
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
150 for k, v in test_label_stats.items():
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
151 if k in exclude:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
152 continue
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
153 if k == "overall_stats":
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
154 continue
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
155 if isinstance(v, (int, float, str, bool)):
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
156 test_metrics[k] = v
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
157
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
158 # 2. Add overall_stats (flattened)
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
159 for k, v in overall_stats.items():
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
160 test_metrics[k] = v
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
161
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
162 # 3. Optionally include combined/loss if present and not already
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
163 if "loss" in combined_stats and "loss" not in test_metrics:
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
164 test_metrics["loss"] = combined_stats["loss"]
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
165 metrics["test"] = test_metrics
bcfa2e234a80 planemo upload for repository https://github.com/goeckslab/gleam.git commit 96bab8325992d16fcaad8e0a4dc4c62b00e2abc2
goeckslab
parents: 11
diff changeset
166 return metrics