Mercurial > repos > goeckslab > pycaret_predict
annotate base_model_trainer.py @ 9:c6c1f8777aae draft default tip
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
author | goeckslab |
---|---|
date | Thu, 31 Jul 2025 15:41:24 +0000 |
parents | 1aed7d47c5ec |
children |
rev | line source |
---|---|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
1 import base64 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
2 import logging |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
3 import tempfile |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
4 from pathlib import Path |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
5 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
6 import h5py |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
7 import joblib |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
8 import numpy as np |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
9 import pandas as pd |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
10 from feature_help_modal import get_feature_metrics_help_modal |
3
ccd798db5abb
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit cf47efb521b91a9cb44ae5c5ade860627f9b9030
goeckslab
parents:
2
diff
changeset
|
11 from feature_importance import FeatureImportanceAnalyzer |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
12 from sklearn.metrics import average_precision_score |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
13 from utils import ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
14 add_hr_to_html, |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
15 add_plot_to_html, |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
16 build_tabbed_html, |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
17 encode_image_to_base64, |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
18 get_html_closing, |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
19 get_html_template, |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
20 ) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
21 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
22 logging.basicConfig(level=logging.DEBUG) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
23 LOG = logging.getLogger(__name__) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
24 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
25 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
26 class BaseModelTrainer: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
27 def __init__( |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
28 self, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
29 input_file, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
30 target_col, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
31 output_dir, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
32 task_type, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
33 random_seed, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
34 test_file=None, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
35 **kwargs, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
36 ): |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
37 self.exp = None |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
38 self.input_file = input_file |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
39 self.target_col = target_col |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
40 self.output_dir = output_dir |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
41 self.task_type = task_type |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
42 self.random_seed = random_seed |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
43 self.data = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
44 self.target = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
45 self.best_model = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
46 self.results = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
47 self.features_name = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
48 self.plots = {} |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
49 self.explainer_plots = {} |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
50 self.plots_explainer_html = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
51 self.trees = [] |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
52 self.user_kwargs = kwargs.copy() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
53 for key, value in self.user_kwargs.items(): |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
54 setattr(self, key, value) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
55 self.setup_params = {} |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
56 self.test_file = test_file |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
57 self.test_data = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
58 |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
59 if not self.output_dir: |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
60 raise ValueError("output_dir must be specified and not None") |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
61 |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
62 LOG.info(f"Model kwargs: {self.__dict__}") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
63 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
64 def load_data(self): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
65 LOG.info(f"Loading data from {self.input_file}") |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
66 self.data = pd.read_csv(self.input_file, sep=None, engine="python") |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
67 self.data.columns = self.data.columns.str.replace(".", "_") |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
68 if "prediction_label" in self.data.columns: |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
69 self.data = self.data.drop(columns=["prediction_label"]) |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
70 |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
71 numeric_cols = self.data.select_dtypes(include=["number"]).columns |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
72 non_numeric_cols = self.data.select_dtypes(exclude=["number"]).columns |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
73 self.data[numeric_cols] = self.data[numeric_cols].apply( |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
74 pd.to_numeric, errors="coerce" |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
75 ) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
76 if len(non_numeric_cols) > 0: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
77 LOG.info(f"Non-numeric columns found: {non_numeric_cols.tolist()}") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
78 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
79 names = self.data.columns.to_list() |
3
ccd798db5abb
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit cf47efb521b91a9cb44ae5c5ade860627f9b9030
goeckslab
parents:
2
diff
changeset
|
80 target_index = int(self.target_col) - 1 |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
81 self.target = names[target_index] |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
82 self.features_name = [n for i, n in enumerate(names) if i != target_index] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
83 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
84 if getattr(self, "missing_value_strategy", None): |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
85 strat = self.missing_value_strategy |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
86 if strat == "mean": |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
87 self.data = self.data.fillna(self.data.mean(numeric_only=True)) |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
88 elif strat == "median": |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
89 self.data = self.data.fillna(self.data.median(numeric_only=True)) |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
90 elif strat == "drop": |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
91 self.data = self.data.dropna() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
92 else: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
93 self.data = self.data.fillna(self.data.median(numeric_only=True)) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
94 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
95 if self.test_file: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
96 LOG.info(f"Loading test data from {self.test_file}") |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
97 df_test = pd.read_csv(self.test_file, sep=None, engine="python") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
98 df_test.columns = df_test.columns.str.replace(".", "_") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
99 self.test_data = df_test |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
100 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
101 def setup_pycaret(self): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
102 LOG.info("Initializing PyCaret") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
103 self.setup_params = { |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
104 "target": self.target, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
105 "session_id": self.random_seed, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
106 "html": True, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
107 "log_experiment": False, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
108 "system_log": False, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
109 "index": False, |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
110 } |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
111 if self.test_data is not None: |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
112 self.setup_params["test_data"] = self.test_data |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
113 for attr in [ |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
114 "train_size", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
115 "normalize", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
116 "feature_selection", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
117 "remove_outliers", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
118 "remove_multicollinearity", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
119 "polynomial_features", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
120 "feature_interaction", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
121 "feature_ratio", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
122 "fix_imbalance", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
123 ]: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
124 val = getattr(self, attr, None) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
125 if val is not None: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
126 self.setup_params[attr] = val |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
127 if getattr(self, "cross_validation_folds", None) is not None: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
128 self.setup_params["fold"] = self.cross_validation_folds |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
129 LOG.info(self.setup_params) |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
130 |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
131 if self.task_type == "classification": |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
132 from pycaret.classification import ClassificationExperiment |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
133 |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
134 self.exp = ClassificationExperiment() |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
135 elif self.task_type == "regression": |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
136 from pycaret.regression import RegressionExperiment |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
137 |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
138 self.exp = RegressionExperiment() |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
139 else: |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
140 raise ValueError("task_type must be 'classification' or 'regression'") |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
141 |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
142 self.exp.setup(self.data, **self.setup_params) |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
143 self.setup_params.update(self.user_kwargs) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
144 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
145 def train_model(self): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
146 LOG.info("Training and selecting the best model") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
147 if self.task_type == "classification": |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
148 self.exp.add_metric( |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
149 id="PR-AUC-Weighted", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
150 name="PR-AUC-Weighted", |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
151 target="pred_proba", |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
152 score_func=average_precision_score, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
153 average="weighted", |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
154 ) |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
155 # Build arguments for compare_models() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
156 compare_kwargs = {} |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
157 if getattr(self, "models", None): |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
158 compare_kwargs["include"] = self.models |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
159 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
160 # Respect explicit cross-validation flag |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
161 if getattr(self, "cross_validation", None) is not None: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
162 compare_kwargs["cross_validation"] = self.cross_validation |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
163 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
164 # Respect explicit fold count |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
165 if getattr(self, "cross_validation_folds", None) is not None: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
166 compare_kwargs["fold"] = self.cross_validation_folds |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
167 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
168 LOG.info(f"compare_models kwargs: {compare_kwargs}") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
169 self.best_model = self.exp.compare_models(**compare_kwargs) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
170 self.results = self.exp.pull() |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
171 if getattr(self, "tune_model", False): |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
172 LOG.info("Tuning hyperparameters of the best model") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
173 self.best_model = self.exp.tune_model(self.best_model) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
174 self.results = self.exp.pull() |
7
f4cb41f458fd
planemo upload for repository https://github.com/goeckslab/gleam commit b430f8b466655878c3bf63b053655fdbf039ddb0
goeckslab
parents:
6
diff
changeset
|
175 |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
176 if self.task_type == "classification": |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
177 self.results.rename(columns={"AUC": "ROC-AUC"}, inplace=True) |
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
178 |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
179 prob_thresh = getattr(self, "probability_threshold", None) |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
180 if self.task_type == "classification" and prob_thresh is not None: |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
181 _ = self.exp.predict_model(self.best_model, probability_threshold=prob_thresh) |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
182 else: |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
183 _ = self.exp.predict_model(self.best_model) |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
184 |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
185 self.test_result_df = self.exp.pull() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
186 if self.task_type == "classification": |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
187 self.test_result_df.rename(columns={"AUC": "ROC-AUC"}, inplace=True) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
188 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
189 def save_model(self): |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
190 hdf5_path = Path(self.output_dir) / "pycaret_model.h5" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
191 with h5py.File(hdf5_path, "w") as f: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
192 with tempfile.NamedTemporaryFile(delete=False) as tmp: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
193 joblib.dump(self.best_model, tmp.name) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
194 tmp.seek(0) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
195 model_bytes = tmp.read() |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
196 f.create_dataset("model", data=np.void(model_bytes)) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
197 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
198 def generate_plots(self): |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
199 LOG.info("Generating PyCaret diagnostic pltos") |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
200 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
201 # choose the right plots based on task |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
202 if self.task_type == "classification": |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
203 plot_names = [ |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
204 "learning", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
205 "vc", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
206 "calibration", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
207 "dimension", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
208 "manifold", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
209 "rfe", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
210 "threshold", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
211 "percentage_above_below", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
212 "class_report", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
213 "pr_auc", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
214 "roc_auc", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
215 ] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
216 else: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
217 plot_names = ["residuals", "vc", "parameter", "error", "learning"] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
218 for name in plot_names: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
219 try: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
220 ax = self.exp.plot_model(self.best_model, plot=name, save=False) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
221 out_path = Path(self.output_dir) / f"plot_{name}.png" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
222 fig = ax.get_figure() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
223 fig.savefig(out_path, bbox_inches="tight") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
224 self.plots[name] = str(out_path) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
225 except Exception as e: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
226 LOG.warning(f"Could not generate {name} plot: {e}") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
227 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
228 def encode_image_to_base64(self, img_path: str) -> str: |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
229 with open(img_path, "rb") as img_file: |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
230 return base64.b64encode(img_file.read()).decode("utf-8") |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
231 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
232 def save_html_report(self): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
233 LOG.info("Saving HTML report") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
234 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
235 # 1) Determine best model name |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
236 try: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
237 best_model_name = str(self.results.iloc[0]["Model"]) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
238 except Exception: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
239 best_model_name = type(self.best_model).__name__ |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
240 LOG.info(f"Best model determined as: {best_model_name}") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
241 |
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
242 # 2) Compute training sample count |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
243 try: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
244 n_train = self.exp.X_train.shape[0] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
245 except Exception: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
246 n_train = getattr(self.exp, "X_train_transformed", pd.DataFrame()).shape[0] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
247 total_rows = self.data.shape[0] |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
248 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
249 # 3) Build setup parameters table |
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
250 all_params = self.setup_params.copy() |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
251 if self.task_type == "classification" and hasattr(self, "probability_threshold"): |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
252 all_params["probability_threshold"] = self.probability_threshold |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
253 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
254 display_keys = [ |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
255 "Target", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
256 "Session ID", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
257 "Train Size", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
258 "Normalize", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
259 "Feature Selection", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
260 "Cross Validation", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
261 "Cross Validation Folds", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
262 "Remove Outliers", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
263 "Remove Multicollinearity", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
264 "Polynomial Features", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
265 "Fix Imbalance", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
266 "Models", |
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
267 "Probability Threshold", |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
268 ] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
269 setup_rows = [] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
270 for key in display_keys: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
271 pk = key.lower().replace(" ", "_") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
272 v = all_params.get(pk) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
273 if key == "Train Size": |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
274 frac = ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
275 float(v) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
276 if v is not None |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
277 else (n_train / total_rows if total_rows else 0) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
278 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
279 dv = f"{frac:.2f} ({n_train} rows)" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
280 elif key in { |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
281 "Normalize", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
282 "Feature Selection", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
283 "Cross Validation", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
284 "Remove Outliers", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
285 "Remove Multicollinearity", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
286 "Polynomial Features", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
287 "Fix Imbalance", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
288 }: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
289 dv = bool(v) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
290 elif key == "Cross Validation Folds": |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
291 dv = v if v is not None else "None" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
292 elif key == "Models": |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
293 dv = ", ".join(map(str, v)) if isinstance(v, (list, tuple)) else "None" |
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
294 elif key == "Probability Threshold": |
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
295 dv = v if v is not None else "None" |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
296 else: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
297 dv = v if v is not None else "None" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
298 setup_rows.append([key, dv]) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
299 if hasattr(self.exp, "_fold_metric"): |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
300 setup_rows.append(["best_model_metric", self.exp._fold_metric]) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
301 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
302 df_setup = pd.DataFrame(setup_rows, columns=["Parameter", "Value"]) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
303 df_setup.to_csv(Path(self.output_dir) / "setup_params.csv", index=False) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
304 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
305 # 4) Persist CSVs |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
306 self.results.to_csv( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
307 Path(self.output_dir) / "comparison_results.csv", index=False |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
308 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
309 self.test_result_df.to_csv( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
310 Path(self.output_dir) / "test_results.csv", index=False |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
311 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
312 pd.DataFrame( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
313 self.best_model.get_params().items(), columns=["Parameter", "Value"] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
314 ).to_csv(Path(self.output_dir) / "best_model.csv", index=False) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
315 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
316 # 5) Header |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
317 header = f"<h2>Best Model: {best_model_name}</h2>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
318 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
319 # — Validation Summary & Configuration — |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
320 val_df = self.results.copy() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
321 # mapping raw plot keys to user-friendly titles |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
322 plot_title_map = { |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
323 "learning": "Learning Curve", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
324 "vc": "Validation Curve", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
325 "calibration": "Calibration Curve", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
326 "dimension": "Dimensionality Reduction", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
327 "manifold": "Manifold Learning", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
328 "rfe": "Recursive Feature Elimination", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
329 "threshold": "Threshold Plot", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
330 "percentage_above_below": "Percentage Above vs. Below Cutoff", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
331 "class_report": "Classification Report", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
332 "pr_auc": "Precision-Recall AUC", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
333 "roc_auc": "Receiver Operating Characteristic AUC", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
334 "residuals": "Residuals Distribution", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
335 "error": "Prediction Error Distribution", |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
336 } |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
337 val_df.drop(columns=["TT (Ec)", "TT (Sec)"], errors="ignore", inplace=True) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
338 summary_html = ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
339 header |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
340 + "<h2>Train & Validation Summary</h2>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
341 + '<div class="table-wrapper">' |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
342 + val_df.to_html(index=False, classes="table sortable") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
343 + "</div>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
344 + "<h2>Setup Parameters</h2>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
345 + '<div class="table-wrapper">' |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
346 + df_setup.to_html(index=False, classes="table sortable") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
347 + "</div>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
348 # — Hyperparameters |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
349 + "<h2>Best Model Hyperparameters</h2>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
350 + '<div class="table-wrapper">' |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
351 + pd.DataFrame( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
352 self.best_model.get_params().items(), columns=["Parameter", "Value"] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
353 ).to_html(index=False, classes="table sortable") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
354 + "</div>" |
3
ccd798db5abb
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit cf47efb521b91a9cb44ae5c5ade860627f9b9030
goeckslab
parents:
2
diff
changeset
|
355 ) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
356 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
357 # choose summary plots based on task type |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
358 if self.task_type == "classification": |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
359 summary_plots = [ |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
360 "learning", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
361 "vc", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
362 "calibration", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
363 "dimension", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
364 "manifold", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
365 "rfe", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
366 "threshold", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
367 "percentage_above_below", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
368 ] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
369 else: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
370 summary_plots = ["learning", "vc", "parameter", "residuals"] |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
371 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
372 for name in summary_plots: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
373 if name in self.plots: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
374 summary_html += "<hr>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
375 b64 = encode_image_to_base64(self.plots[name]) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
376 title = plot_title_map.get(name, name.replace("_", " ").title()) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
377 summary_html += ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
378 '<div class="plot">' |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
379 f"<h2>{title}</h2>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
380 f'<img src="data:image/png;base64,{b64}" ' |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
381 'style="max-width:90%;max-height:600px;border:1px solid #ddd;"/>' |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
382 "</div>" |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
383 ) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
384 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
385 # — Test Summary — |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
386 test_html = ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
387 header |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
388 + '<div class="table-wrapper">' |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
389 + self.test_result_df.to_html(index=False, classes="table sortable") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
390 + "</div>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
391 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
392 if self.task_type == "regression": |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
393 try: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
394 y_true = ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
395 pd.Series(self.exp.y_test_transformed) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
396 .reset_index(drop=True) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
397 .rename("True") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
398 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
399 y_pred = pd.Series( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
400 self.best_model.predict(self.exp.X_test_transformed) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
401 ).rename("Predicted") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
402 df_tp = pd.concat([y_true, y_pred], axis=1) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
403 test_html += "<h2>True vs Predicted Values</h2>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
404 test_html += ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
405 '<div class="table-wrapper" style="max-height:400px; overflow-y:auto;">' |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
406 + df_tp.head(50).to_html(index=False, classes="table sortable") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
407 + "</div>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
408 + add_hr_to_html() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
409 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
410 except Exception as e: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
411 LOG.warning(f"Could not generate True vs Predicted table: {e}") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
412 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
413 # 5a) Explainer-substituted plots in order |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
414 if self.task_type == "regression": |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
415 test_order = ["residuals"] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
416 else: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
417 test_order = [ |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
418 "confusion_matrix", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
419 "roc_auc", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
420 "pr_auc", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
421 "lift_curve", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
422 "threshold", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
423 "cumulative_precision", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
424 ] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
425 for key in test_order: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
426 fig_or_fn = self.explainer_plots.pop(key, None) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
427 if fig_or_fn is not None: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
428 fig = fig_or_fn() if callable(fig_or_fn) else fig_or_fn |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
429 title = plot_title_map.get(key, key.replace("_", " ").title()) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
430 test_html += ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
431 f"<h2>{title}</h2>" + add_plot_to_html(fig) + add_hr_to_html() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
432 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
433 # 5b) Remaining PyCaret test plots |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
434 for name, path in self.plots.items(): |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
435 # classification: include only the small extras, before skipping anything |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
436 if self.task_type == "classification" and name in { |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
437 "threshold", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
438 "pr_auc", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
439 "class_report", |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
440 }: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
441 title = plot_title_map.get(name, name.replace("_", " ").title()) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
442 b64 = encode_image_to_base64(path) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
443 test_html += ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
444 f"<h2>{title}</h2>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
445 "<div class='plot'>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
446 f"<img src='data:image/png;base64,{b64}' " |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
447 "style='max-width:90%;max-height:600px;border:1px solid #ddd;'/>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
448 "</div>" + add_hr_to_html() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
449 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
450 continue |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
451 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
452 # regression: explicitly include the 'error' plot, before skipping |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
453 if self.task_type == "regression" and name == "error": |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
454 title = plot_title_map.get("error", "Prediction Error Distribution") |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
455 b64 = encode_image_to_base64(path) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
456 test_html += ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
457 f"<h2>{title}</h2>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
458 "<div class='plot'>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
459 f"<img src='data:image/png;base64,{b64}' " |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
460 "style='max-width:90%;max-height:600px;border:1px solid #ddd;'/>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
461 "</div>" + add_hr_to_html() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
462 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
463 continue |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
464 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
465 # now skip any plots already rendered via test_order |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
466 if name in test_order: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
467 continue |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
468 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
469 # — Feature Importance — |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
470 feature_html = header |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
471 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
472 # 6a) PyCaret’s default feature importances |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
473 feature_html += FeatureImportanceAnalyzer( |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
474 data=self.data, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
475 target_col=self.target_col, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
476 task_type=self.task_type, |
3
ccd798db5abb
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit cf47efb521b91a9cb44ae5c5ade860627f9b9030
goeckslab
parents:
2
diff
changeset
|
477 output_dir=self.output_dir, |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
478 exp=self.exp, |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
479 best_model=self.best_model, |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
480 ).run() |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
481 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
482 # 6b) Explainer SHAP importances |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
483 for key in ["shap_mean", "shap_perm"]: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
484 fig_or_fn = self.explainer_plots.pop(key, None) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
485 if fig_or_fn is not None: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
486 fig = fig_or_fn() if callable(fig_or_fn) else fig_or_fn |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
487 # give SHAP plots explicit titles |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
488 title = ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
489 "Mean Absolute SHAP Value Impact" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
490 if key == "shap_mean" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
491 else "Permutation Feature Importance" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
492 ) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
493 feature_html += ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
494 f"<h2>{title}</h2>" + add_plot_to_html(fig) + add_hr_to_html() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
495 ) |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
496 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
497 # 6c) PDPs last |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
498 pdp_keys = sorted(k for k in self.explainer_plots if k.startswith("pdp__")) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
499 for k in pdp_keys: |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
500 fig_or_fn = self.explainer_plots[k] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
501 fig = fig_or_fn() if callable(fig_or_fn) else fig_or_fn |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
502 # extract feature name |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
503 feature = k.split("__", 1)[1] |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
504 title = f"Partial Dependence for {feature}" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
505 feature_html += ( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
506 f"<h2>{title}</h2>" + add_plot_to_html(fig) + add_hr_to_html() |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
507 ) |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
508 # 7) Assemble final HTML (three tabs) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
509 html = get_html_template() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
510 html += "<h1>Tabular Learner Model Report</h1>" |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
511 html += build_tabbed_html(summary_html, test_html, feature_html) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
512 html += get_feature_metrics_help_modal() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
513 html += get_html_closing() |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
514 |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
515 # 8) Write out |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
516 (Path(self.output_dir) / "comparison_result.html").write_text( |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
517 html, encoding="utf-8" |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
518 ) |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
519 LOG.info(f"HTML report generated at: {self.output_dir}/comparison_result.html") |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
520 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
521 def save_dashboard(self): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
522 raise NotImplementedError("Subclasses should implement this method") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
523 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
524 def generate_plots_explainer(self): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
525 raise NotImplementedError("Subclasses should implement this method") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
526 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
527 def generate_tree_plots(self): |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
528 from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
529 from xgboost import XGBClassifier, XGBRegressor |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
530 from explainerdashboard.explainers import RandomForestExplainer |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
531 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
532 LOG.info("Generating tree plots") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
533 X_test = self.exp.X_test_transformed.copy() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
534 y_test = self.exp.y_test_transformed |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
535 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
536 if isinstance(self.best_model, (RandomForestClassifier, RandomForestRegressor)): |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
537 n_trees = self.best_model.n_estimators |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
538 elif isinstance(self.best_model, (XGBClassifier, XGBRegressor)): |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
539 n_trees = len(self.best_model.get_booster().get_dump()) |
6
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
540 else: |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
541 LOG.warning("Tree plots not supported for this model type.") |
a32ff7201629
planemo upload for repository https://github.com/goeckslab/gleam commit 06c0da44ac93256dfb616a6b40276b5485a71e8e
goeckslab
parents:
3
diff
changeset
|
542 return |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
543 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
544 explainer = RandomForestExplainer(self.best_model, X_test, y_test) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
545 for i in range(n_trees): |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
546 fig = explainer.decisiontree_encoded(tree_idx=i, index=0) |
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
547 self.trees.append(fig) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
548 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
549 def run(self): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
550 self.load_data() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
551 self.setup_pycaret() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
552 self.train_model() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
553 self.save_model() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
554 self.generate_plots() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
555 self.generate_plots_explainer() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
556 self.generate_tree_plots() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
557 self.save_html_report() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
558 # self.save_dashboard() |