Mercurial > repos > goeckslab > pycaret_predict
annotate pycaret_train.py @ 13:f07850192bc2 draft default tip
planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
| author | goeckslab |
|---|---|
| date | Sat, 08 Nov 2025 14:20:33 +0000 |
| parents | e2a6fed32d54 |
| 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 |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
4 from pycaret_classification import ClassificationModelTrainer |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
5 from pycaret_regression import RegressionModelTrainer |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
6 |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
7 logging.basicConfig(level=logging.DEBUG) |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
8 LOG = logging.getLogger(__name__) |
|
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 |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
11 def main(): |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
12 parser = argparse.ArgumentParser() |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
13 parser.add_argument("--input_file", help="Path to the input file") |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
14 parser.add_argument("--target_col", help="Column number of the target") |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
15 parser.add_argument("--output_dir", help="Path to the output directory") |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
16 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
17 "--model_type", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
18 choices=["classification", "regression"], |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
19 help="Type of the model", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
20 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
21 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
22 "--train_size", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
23 type=float, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
24 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
25 help="Train size for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
26 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
27 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
28 "--normalize", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
29 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
30 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
31 help="Normalize data for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
32 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
33 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
34 "--feature_selection", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
35 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
36 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
37 help="Perform feature selection for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
38 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
39 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
40 "--cross_validation", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
41 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
42 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
43 help="Enable cross-validation for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
44 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
45 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
46 "--no_cross_validation", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
47 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
48 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
49 help="Disable cross-validation for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
50 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
51 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
52 "--cross_validation_folds", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
53 type=int, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
54 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
55 help="Number of cross-validation folds for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
56 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
57 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
58 "--remove_outliers", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
59 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
60 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
61 help="Remove outliers for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
62 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
63 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
64 "--remove_multicollinearity", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
65 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
66 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
67 help="Remove multicollinearity for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
68 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
69 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
70 "--polynomial_features", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
71 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
72 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
73 help="Generate polynomial features for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
74 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
75 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
76 "--feature_interaction", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
77 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
78 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
79 help="Generate feature interactions for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
80 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
81 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
82 "--feature_ratio", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
83 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
84 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
85 help="Generate feature ratios for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
86 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
87 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
88 "--fix_imbalance", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
89 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
90 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
91 help="Fix class imbalance for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
92 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
93 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
94 "--models", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
95 nargs="+", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
96 default=None, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
97 help="Selected models for training", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
98 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
99 parser.add_argument( |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
100 "--tune_model", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
101 action="store_true", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
102 default=False, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
103 help="Tune the best model hyperparameters after training", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
104 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
105 parser.add_argument( |
|
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
106 "--test_file", |
|
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
107 type=str, |
|
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
108 default=None, |
|
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
109 help="Path to the test data file", |
|
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
110 ) |
|
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
111 parser.add_argument( |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
112 "--random_seed", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
113 type=int, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
114 default=42, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
115 help="Random seed for PyCaret setup", |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
116 ) |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
117 parser.add_argument( |
|
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
118 "--probability_threshold", |
|
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
119 type=float, |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
120 default=None, |
|
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
121 help="Probability threshold for classification decision,", |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
122 ) |
|
13
f07850192bc2
planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents:
10
diff
changeset
|
123 parser.add_argument( |
|
f07850192bc2
planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents:
10
diff
changeset
|
124 "--best_model_metric", |
|
f07850192bc2
planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents:
10
diff
changeset
|
125 type=str, |
|
f07850192bc2
planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents:
10
diff
changeset
|
126 default=None, |
|
f07850192bc2
planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents:
10
diff
changeset
|
127 help="Metric used to select the best model (e.g. AUC, Accuracy, R2, RMSE).", |
|
f07850192bc2
planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents:
10
diff
changeset
|
128 ) |
|
0
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 args = parser.parse_args() |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
131 |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
132 # Normalize cross-validation flags: --no_cross_validation overrides --cross_validation |
|
7
f4cb41f458fd
planemo upload for repository https://github.com/goeckslab/gleam commit b430f8b466655878c3bf63b053655fdbf039ddb0
goeckslab
parents:
3
diff
changeset
|
133 if args.no_cross_validation: |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
134 args.cross_validation = False |
|
10
e2a6fed32d54
planemo upload for repository https://github.com/goeckslab/gleam commit 47a5977e074223e92e216efa42969a4056516707
goeckslab
parents:
9
diff
changeset
|
135 # If --cross_validation was passed, args.cross_validation is True |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
136 # If neither was passed, args.cross_validation remains None |
|
7
f4cb41f458fd
planemo upload for repository https://github.com/goeckslab/gleam commit b430f8b466655878c3bf63b053655fdbf039ddb0
goeckslab
parents:
3
diff
changeset
|
137 |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
138 # Build the model_kwargs dict from CLI args |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
139 model_kwargs = { |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
140 "train_size": args.train_size, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
141 "normalize": args.normalize, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
142 "feature_selection": args.feature_selection, |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
143 "cross_validation": args.cross_validation, |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
144 "cross_validation_folds": args.cross_validation_folds, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
145 "remove_outliers": args.remove_outliers, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
146 "remove_multicollinearity": args.remove_multicollinearity, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
147 "polynomial_features": args.polynomial_features, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
148 "feature_interaction": args.feature_interaction, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
149 "feature_ratio": args.feature_ratio, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
150 "fix_imbalance": args.fix_imbalance, |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
151 "tune_model": args.tune_model, |
|
9
c6c1f8777aae
planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents:
8
diff
changeset
|
152 "probability_threshold": args.probability_threshold, |
|
13
f07850192bc2
planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents:
10
diff
changeset
|
153 "best_model_metric": args.best_model_metric, |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
154 } |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
155 LOG.info(f"Model kwargs: {model_kwargs}") |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
156 |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
157 # If the XML passed a comma-separated string in a single list element, split it out |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
158 if args.models: |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
159 model_kwargs["models"] = args.models[0].split(",") |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
160 |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
161 # Drop None entries so PyCaret uses its default values |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
162 model_kwargs = {k: v for k, v in model_kwargs.items() if v is not None} |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
163 LOG.info(f"Model kwargs 2: {model_kwargs}") |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
164 |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
165 # Instantiate the appropriate trainer |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
166 if args.model_type == "classification": |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
167 trainer = ClassificationModelTrainer( |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
168 args.input_file, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
169 args.target_col, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
170 args.output_dir, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
171 args.model_type, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
172 args.random_seed, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
173 args.test_file, |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
174 **model_kwargs, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
175 ) |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
176 elif args.model_type == "regression": |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
177 # regression doesn't support fix_imbalance |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
178 model_kwargs.pop("fix_imbalance", None) |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
179 trainer = RegressionModelTrainer( |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
180 args.input_file, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
181 args.target_col, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
182 args.output_dir, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
183 args.model_type, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
184 args.random_seed, |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
185 args.test_file, |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
186 **model_kwargs, |
|
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
187 ) |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
188 else: |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
189 LOG.error("Invalid model type. Please choose 'classification' or 'regression'.") |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
190 return |
|
8
1aed7d47c5ec
planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents:
7
diff
changeset
|
191 |
|
0
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
192 trainer.run() |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
193 |
|
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 if __name__ == "__main__": |
|
1f20fe57fdee
planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff
changeset
|
196 main() |
