annotate feature_importance.py @ 3:ccd798db5abb draft default tip

planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit cf47efb521b91a9cb44ae5c5ade860627f9b9030
author goeckslab
date Tue, 03 Jun 2025 19:31:06 +0000
parents 1f20fe57fdee
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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 os
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
4
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
5 import matplotlib.pyplot as plt
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
6 import pandas as pd
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
7 from pycaret.classification import ClassificationExperiment
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
8 from pycaret.regression import RegressionExperiment
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
9
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
10 logging.basicConfig(level=logging.DEBUG)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
11 LOG = logging.getLogger(__name__)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
12
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
13
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
14 class FeatureImportanceAnalyzer:
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
15 def __init__(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
16 self,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
17 task_type,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
18 output_dir,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
19 data_path=None,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
20 data=None,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
21 target_col=None):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
22
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
23 if data is not None:
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
24 self.data = data
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
25 LOG.info("Data loaded from memory")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
26 else:
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
27 self.target_col = target_col
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
28 self.data = pd.read_csv(data_path, sep=None, engine='python')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
29 self.data.columns = self.data.columns.str.replace('.', '_')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
30 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
31 self.task_type = task_type
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
32 self.target = self.data.columns[int(target_col) - 1]
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
33 self.exp = ClassificationExperiment() \
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
34 if task_type == 'classification' \
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
35 else RegressionExperiment()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
36 self.plots = {}
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
37 self.output_dir = output_dir
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
38
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
39 def setup_pycaret(self):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
40 LOG.info("Initializing PyCaret")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
41 setup_params = {
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
42 'target': self.target,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
43 'session_id': 123,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
44 'html': True,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
45 'log_experiment': False,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
46 'system_log': False
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
47 }
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
48 LOG.info(self.task_type)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
49 LOG.info(self.exp)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
50 self.exp.setup(self.data, **setup_params)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
51
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
52 # def save_coefficients(self):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
53 # model = self.exp.create_model('lr')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
54 # coef_df = pd.DataFrame({
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
55 # 'Feature': self.data.columns.drop(self.target),
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
56 # 'Coefficient': model.coef_[0]
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
57 # })
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
58 # coef_html = coef_df.to_html(index=False)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
59 # return coef_html
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
60
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
61 def save_tree_importance(self):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
62 model = self.exp.create_model('rf')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
63 importances = model.feature_importances_
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
64 processed_features = self.exp.get_config('X_transformed').columns
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
65 LOG.debug(f"Feature importances: {importances}")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
66 LOG.debug(f"Features: {processed_features}")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
67 feature_importances = pd.DataFrame({
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
68 'Feature': processed_features,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
69 'Importance': importances
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
70 }).sort_values(by='Importance', ascending=False)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
71 plt.figure(figsize=(10, 6))
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
72 plt.barh(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
73 feature_importances['Feature'],
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
74 feature_importances['Importance'])
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
75 plt.xlabel('Importance')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
76 plt.title('Feature Importance (Random Forest)')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
77 plot_path = os.path.join(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
78 self.output_dir,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
79 'tree_importance.png')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
80 plt.savefig(plot_path)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
81 plt.close()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
82 self.plots['tree_importance'] = plot_path
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
83
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
84 def save_shap_values(self):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
85 model = self.exp.create_model('lightgbm')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
86 import shap
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
87 explainer = shap.Explainer(model)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
88 shap_values = explainer.shap_values(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
89 self.exp.get_config('X_transformed'))
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
90 shap.summary_plot(shap_values,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
91 self.exp.get_config('X_transformed'), show=False)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
92 plt.title('Shap (LightGBM)')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
93 plot_path = os.path.join(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
94 self.output_dir, 'shap_summary.png')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
95 plt.savefig(plot_path)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
96 plt.close()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
97 self.plots['shap_summary'] = plot_path
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
98
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
99 def generate_feature_importance(self):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
100 # coef_html = self.save_coefficients()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
101 self.save_tree_importance()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
102 self.save_shap_values()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
103
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
104 def encode_image_to_base64(self, img_path):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
105 with open(img_path, 'rb') as img_file:
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
106 return base64.b64encode(img_file.read()).decode('utf-8')
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
107
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
108 def generate_html_report(self):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
109 LOG.info("Generating HTML report")
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 # Read and encode plot images
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
112 plots_html = ""
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
113 for plot_name, plot_path in self.plots.items():
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
114 encoded_image = self.encode_image_to_base64(plot_path)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
115 plots_html += f"""
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
116 <div class="plot" id="{plot_name}">
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
117 <h2>{'Feature importance analysis from a'
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
118 'trained Random Forest'
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
119 if plot_name == 'tree_importance'
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
120 else 'SHAP Summary from a trained lightgbm'}</h2>
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
121 <h3>{'Use gini impurity for'
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
122 'calculating feature importance for classification'
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
123 'and Variance Reduction for regression'
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
124 if plot_name == 'tree_importance'
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
125 else ''}</h3>
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
126 <img src="data:image/png;base64,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
127 {encoded_image}" alt="{plot_name}">
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
128 </div>
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
129 """
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
130
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
131 # Generate HTML content with tabs
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
132 html_content = f"""
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
133 <h1>PyCaret Feature Importance Report</h1>
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
134 {plots_html}
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
135 """
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
136
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
137 return html_content
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
138
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
139 def run(self):
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
140 LOG.info("Running feature importance analysis")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
141 self.setup_pycaret()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
142 self.generate_feature_importance()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
143 html_content = self.generate_html_report()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
144 LOG.info("Feature importance analysis completed")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
145 return html_content
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
146
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
147
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
148 if __name__ == "__main__":
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
149 import argparse
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
150 parser = argparse.ArgumentParser(description="Feature Importance Analysis")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
151 parser.add_argument(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
152 "--data_path", type=str, help="Path to the dataset")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
153 parser.add_argument(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
154 "--target_col", type=int,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
155 help="Index of the target column (1-based)")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
156 parser.add_argument(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
157 "--task_type", type=str,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
158 choices=["classification", "regression"],
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
159 help="Task type: classification or regression")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
160 parser.add_argument(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
161 "--output_dir",
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
162 type=str,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
163 help="Directory to save the outputs")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
164 args = parser.parse_args()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
165
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
166 analyzer = FeatureImportanceAnalyzer(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
167 args.data_path, args.target_col,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
168 args.task_type, args.output_dir)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
169 analyzer.run()