annotate pycaret_train.py @ 14:7d78a6afc958 draft default tip

planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
author goeckslab
date Wed, 26 Nov 2025 17:49:50 +0000
parents f07850192bc2
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 argparse
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
2 import logging
14
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
3 import os
0
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 from pycaret_classification import ClassificationModelTrainer
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
6 from pycaret_regression import RegressionModelTrainer
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
7
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
8 logging.basicConfig(level=logging.DEBUG)
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
9 LOG = logging.getLogger(__name__)
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
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
12 def main():
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
13 parser = argparse.ArgumentParser()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
14 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
15 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
16 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
17 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
18 "--model_type",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
19 choices=["classification", "regression"],
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
20 help="Type of the model",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
21 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
22 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
23 "--train_size",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
24 type=float,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
25 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
26 help="Train size for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
27 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
28 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
29 "--normalize",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
30 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
31 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
32 help="Normalize data for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
33 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
34 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
35 "--feature_selection",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
36 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
37 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
38 help="Perform feature selection for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
39 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
40 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
41 "--cross_validation",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
42 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
43 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
44 help="Enable cross-validation for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
45 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
46 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
47 "--no_cross_validation",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
48 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
49 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
50 help="Disable cross-validation for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
51 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
52 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
53 "--cross_validation_folds",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
54 type=int,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
55 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
56 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
57 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
58 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
59 "--remove_outliers",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
60 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
61 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
62 help="Remove outliers for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
63 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
64 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
65 "--remove_multicollinearity",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
66 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
67 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
68 help="Remove multicollinearity for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
69 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
70 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
71 "--polynomial_features",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
72 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
73 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
74 help="Generate polynomial features for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
75 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
76 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
77 "--feature_interaction",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
78 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
79 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
80 help="Generate feature interactions for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
81 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
82 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
83 "--feature_ratio",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
84 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
85 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
86 help="Generate feature ratios for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
87 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
88 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
89 "--fix_imbalance",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
90 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
91 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
92 help="Fix class imbalance for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
93 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
94 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
95 "--models",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
96 nargs="+",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
97 default=None,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
98 help="Selected models for training",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
99 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
100 parser.add_argument(
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
101 "--tune_model",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
102 action="store_true",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
103 default=False,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
104 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
105 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
106 parser.add_argument(
9
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
107 "--test_file",
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
108 type=str,
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
109 default=None,
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
110 help="Path to the test data file",
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
111 )
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
112 parser.add_argument(
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
113 "--random_seed",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
114 type=int,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
115 default=42,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
116 help="Random seed for PyCaret setup",
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
117 )
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
118 parser.add_argument(
14
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
119 "--n-jobs",
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
120 dest="n_jobs",
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
121 type=int,
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
122 default=None,
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
123 help="Number of parallel jobs; defaults to GALAXY_SLOTS or 1 if unset/invalid.",
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
124 )
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
125 parser.add_argument(
9
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
126 "--probability_threshold",
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
127 type=float,
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
128 default=None,
9
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
129 help="Probability threshold for classification decision,",
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
130 )
13
f07850192bc2 planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents: 10
diff changeset
131 parser.add_argument(
f07850192bc2 planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents: 10
diff changeset
132 "--best_model_metric",
f07850192bc2 planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents: 10
diff changeset
133 type=str,
f07850192bc2 planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents: 10
diff changeset
134 default=None,
f07850192bc2 planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents: 10
diff changeset
135 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
136 )
0
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 args = parser.parse_args()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
139
14
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
140 # Derive n_jobs from CLI or GALAXY_SLOTS env var
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
141 if args.n_jobs is not None:
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
142 n_jobs = args.n_jobs
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
143 else:
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
144 slots_str = os.environ.get("GALAXY_SLOTS")
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
145 try:
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
146 n_jobs = int(slots_str) if slots_str is not None else 1
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
147 except ValueError:
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
148 n_jobs = 1
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
149
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
150 # 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
151 if args.no_cross_validation:
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
152 args.cross_validation = False
10
e2a6fed32d54 planemo upload for repository https://github.com/goeckslab/gleam commit 47a5977e074223e92e216efa42969a4056516707
goeckslab
parents: 9
diff changeset
153 # 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
154 # 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
155
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
156 # 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
157 model_kwargs = {
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
158 "train_size": args.train_size,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
159 "normalize": args.normalize,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
160 "feature_selection": args.feature_selection,
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
161 "cross_validation": args.cross_validation,
0
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
162 "cross_validation_folds": args.cross_validation_folds,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
163 "remove_outliers": args.remove_outliers,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
164 "remove_multicollinearity": args.remove_multicollinearity,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
165 "polynomial_features": args.polynomial_features,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
166 "feature_interaction": args.feature_interaction,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
167 "feature_ratio": args.feature_ratio,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
168 "fix_imbalance": args.fix_imbalance,
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
169 "tune_model": args.tune_model,
14
7d78a6afc958 planemo upload for repository https://github.com/goeckslab/gleam commit 1ffd143e57fa952ee9dd84fc141771520aea0791
goeckslab
parents: 13
diff changeset
170 "n_jobs": n_jobs,
9
c6c1f8777aae planemo upload for repository https://github.com/goeckslab/gleam commit 4a11e8a4c4e9daa884bddedfa47090476c517667
goeckslab
parents: 8
diff changeset
171 "probability_threshold": args.probability_threshold,
13
f07850192bc2 planemo upload for repository https://github.com/goeckslab/gleam commit 84d5cd0b1fa5c1ff0ad892bc39c95dad1ceb4920
goeckslab
parents: 10
diff changeset
172 "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
173 }
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
174 LOG.info(f"Model kwargs: {model_kwargs}")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
175
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
176 # 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
177 if args.models:
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
178 model_kwargs["models"] = args.models[0].split(",")
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
179
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
180 # 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
181 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
182 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
183
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
184 # Instantiate the appropriate trainer
0
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
185 if args.model_type == "classification":
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
186 trainer = ClassificationModelTrainer(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
187 args.input_file,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
188 args.target_col,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
189 args.output_dir,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
190 args.model_type,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
191 args.random_seed,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
192 args.test_file,
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
193 **model_kwargs,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
194 )
0
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
195 elif args.model_type == "regression":
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
196 # regression doesn't support fix_imbalance
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
197 model_kwargs.pop("fix_imbalance", None)
0
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
198 trainer = RegressionModelTrainer(
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
199 args.input_file,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
200 args.target_col,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
201 args.output_dir,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
202 args.model_type,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
203 args.random_seed,
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
204 args.test_file,
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
205 **model_kwargs,
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
206 )
0
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
207 else:
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
208 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
209 return
8
1aed7d47c5ec planemo upload for repository https://github.com/goeckslab/gleam commit 8112548ac44b7a4769093d76c722c8fcdeaaef54
goeckslab
parents: 7
diff changeset
210
0
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
211 trainer.run()
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
212
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
213
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
214 if __name__ == "__main__":
1f20fe57fdee planemo upload for repository https://github.com/goeckslab/Galaxy-Pycaret commit d79b0f722b7d09505a526d1a4332f87e548a3df1
goeckslab
parents:
diff changeset
215 main()