Mercurial > repos > goeckslab > pycaret_predict
annotate pycaret_predict.py @ 8:1aed7d47c5ec draft default tip
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
author | goeckslab |
---|---|
date | Fri, 25 Jul 2025 19:02:32 +0000 |
parents | ccd798db5abb |
children |
rev | line source |
---|---|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
1 import argparse |
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 |
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 h5py |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
6 import joblib |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
7 import pandas as pd |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
8 from pycaret.classification import ClassificationExperiment |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
9 from pycaret.regression import RegressionExperiment |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
10 from sklearn.metrics import average_precision_score |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
11 from utils import encode_image_to_base64, get_html_closing, get_html_template |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
12 |
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
3
diff
changeset
|
13 |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
14 LOG = logging.getLogger(__name__) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
15 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
16 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
17 class PyCaretModelEvaluator: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
18 def __init__(self, model_path, task, target): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
19 self.model_path = model_path |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
20 self.task = task.lower() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
21 self.model = self.load_h5_model() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
22 self.target = target if target != "None" else None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
23 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
24 def load_h5_model(self): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
25 """Load a PyCaret model from an HDF5 file.""" |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
26 with h5py.File(self.model_path, 'r') as f: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
27 model_bytes = bytes(f['model'][()]) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
28 with tempfile.NamedTemporaryFile(delete=False) as temp_file: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
29 temp_file.write(model_bytes) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
30 temp_file.seek(0) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
31 loaded_model = joblib.load(temp_file.name) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
32 return loaded_model |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
33 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
34 def evaluate(self, data_path): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
35 """Evaluate the model using the specified data.""" |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
36 raise NotImplementedError("Subclasses must implement this method") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
37 |
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 class ClassificationEvaluator(PyCaretModelEvaluator): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
40 def evaluate(self, data_path): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
41 metrics = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
42 plot_paths = {} |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
43 data = pd.read_csv(data_path, engine='python', sep=None) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
44 if self.target: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
45 exp = ClassificationExperiment() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
46 names = data.columns.to_list() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
47 LOG.error(f"Column names: {names}") |
3
ccd798db5abb
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit cf47efb521b91a9cb44ae5c5ade860627f9b9030
goeckslab
parents:
0
diff
changeset
|
48 target_index = int(self.target) - 1 |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
49 target_name = names[target_index] |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
50 exp.setup(data, target=target_name, test_data=data, index=False) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
51 exp.add_metric(id='PR-AUC-Weighted', |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
52 name='PR-AUC-Weighted', |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
53 target='pred_proba', |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
54 score_func=average_precision_score, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
55 average='weighted') |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
56 predictions = exp.predict_model(self.model) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
57 metrics = exp.pull() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
58 plots = ['confusion_matrix', 'auc', 'threshold', 'pr', |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
59 'error', 'class_report', 'learning', 'calibration', |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
60 'vc', 'dimension', 'manifold', 'rfe', 'feature', |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
61 'feature_all'] |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
62 for plot_name in plots: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
63 try: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
64 if plot_name == 'auc' and not exp.is_multiclass: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
65 plot_path = exp.plot_model(self.model, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
66 plot=plot_name, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
67 save=True, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
68 plot_kwargs={ |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
69 'micro': False, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
70 'macro': False, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
71 'per_class': False, |
3
ccd798db5abb
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit cf47efb521b91a9cb44ae5c5ade860627f9b9030
goeckslab
parents:
0
diff
changeset
|
72 'binary': True}) |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
73 plot_paths[plot_name] = plot_path |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
74 continue |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
75 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
76 plot_path = exp.plot_model(self.model, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
77 plot=plot_name, save=True) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
78 plot_paths[plot_name] = plot_path |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
79 except Exception as e: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
80 LOG.error(f"Error generating plot {plot_name}: {e}") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
81 continue |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
82 generate_html_report(plot_paths, metrics) |
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 else: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
85 exp = ClassificationExperiment() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
86 exp.setup(data, target=None, test_data=data, index=False) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
87 predictions = exp.predict_model(self.model, data=data) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
88 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
89 return predictions, metrics, plot_paths |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
90 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
91 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
92 class RegressionEvaluator(PyCaretModelEvaluator): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
93 def evaluate(self, data_path): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
94 metrics = None |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
95 plot_paths = {} |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
96 data = pd.read_csv(data_path, engine='python', sep=None) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
97 if self.target: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
98 names = data.columns.to_list() |
3
ccd798db5abb
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit cf47efb521b91a9cb44ae5c5ade860627f9b9030
goeckslab
parents:
0
diff
changeset
|
99 target_index = int(self.target) - 1 |
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
100 target_name = names[target_index] |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
101 exp = RegressionExperiment() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
102 exp.setup(data, target=target_name, test_data=data, index=False) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
103 predictions = exp.predict_model(self.model) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
104 metrics = exp.pull() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
105 plots = ['residuals', 'error', 'cooks', |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
106 'learning', 'vc', 'manifold', |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
107 'rfe', 'feature', 'feature_all'] |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
108 for plot_name in plots: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
109 try: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
110 plot_path = exp.plot_model(self.model, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
111 plot=plot_name, save=True) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
112 plot_paths[plot_name] = plot_path |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
113 except Exception as e: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
114 LOG.error(f"Error generating plot {plot_name}: {e}") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
115 continue |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
116 generate_html_report(plot_paths, metrics) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
117 else: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
118 exp = RegressionExperiment() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
119 exp.setup(data, target=None, test_data=data, index=False) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
120 predictions = exp.predict_model(self.model, data=data) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
121 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
122 return predictions, metrics, plot_paths |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
123 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
124 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
125 def generate_html_report(plots, metrics): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
126 """Generate an HTML evaluation report.""" |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
127 plots_html = "" |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
128 for plot_name, plot_path in plots.items(): |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
129 encoded_image = encode_image_to_base64(plot_path) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
130 plots_html += f""" |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
131 <div class="plot"> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
132 <h3>{plot_name.capitalize()}</h3> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
133 <img src="data:image/png;base64,{encoded_image}" alt="{plot_name}"> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
134 </div> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
135 <hr> |
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 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
138 metrics_html = metrics.to_html(index=False, classes="table") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
139 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
140 html_content = f""" |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
141 {get_html_template()} |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
142 <h1>Model Evaluation Report</h1> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
143 <div class="tabs"> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
144 <div class="tab" onclick="openTab(event, 'metrics')">Metrics</div> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
145 <div class="tab" onclick="openTab(event, 'plots')">Plots</div> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
146 </div> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
147 <div id="metrics" class="tab-content"> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
148 <h2>Metrics</h2> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
149 <table> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
150 {metrics_html} |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
151 </table> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
152 </div> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
153 <div id="plots" class="tab-content"> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
154 <h2>Plots</h2> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
155 {plots_html} |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
156 </div> |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
157 {get_html_closing()} |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
158 """ |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
159 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
160 # Save HTML report |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
161 with open("evaluation_report.html", "w") as html_file: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
162 html_file.write(html_content) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
163 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
164 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
165 if __name__ == "__main__": |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
166 parser = argparse.ArgumentParser( |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
167 description="Evaluate a PyCaret model stored in HDF5 format.") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
168 parser.add_argument("--model_path", |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
169 type=str, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
170 help="Path to the HDF5 model file.") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
171 parser.add_argument("--data_path", |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
172 type=str, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
173 help="Path to the evaluation data CSV file.") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
174 parser.add_argument("--task", |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
175 type=str, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
176 choices=["classification", "regression"], |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
177 help="Specify the task: classification or regression.") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
178 parser.add_argument("--target", |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
179 default=None, |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
180 help="Column number of the target") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
181 args = parser.parse_args() |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
182 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
183 if args.task == "classification": |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
184 evaluator = ClassificationEvaluator( |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
185 args.model_path, args.task, args.target) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
186 elif args.task == "regression": |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
187 evaluator = RegressionEvaluator( |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
188 args.model_path, args.task, args.target) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
189 else: |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
190 raise ValueError( |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
191 "Unsupported task type. Use 'classification' or 'regression'.") |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
192 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
193 predictions, metrics, plots = evaluator.evaluate(args.data_path) |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
194 |
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
195 predictions.to_csv("predictions.csv", index=False) |