Mercurial > repos > bgruening > sklearn_feature_selection

annotate train_test_split.py @ 34:b9d86fc6359d draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
author bgruening
date Wed, 09 Aug 2023 11:58:39 +0000
parents 75e9f6fb6536
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
1 import argparse
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
2 import json
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
3 import warnings
34
b9d86fc6359d planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 29
diff changeset
4 from distutils.version import LooseVersion as Version
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
5
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
6 import pandas as pd
34
b9d86fc6359d planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 29
diff changeset
7 from galaxy_ml import __version__ as galaxy_ml_version
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
8 from galaxy_ml.model_validations import train_test_split
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
9 from galaxy_ml.utils import get_cv, read_columns
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
10
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
11
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
12 def _get_single_cv_split(params, array, infile_labels=None, infile_groups=None):
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
13 """output (train, test) subset from a cv splitter
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
14
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
15 Parameters
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
16 ----------
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
17 params : dict
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
18 Galaxy tool inputs
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
19 array : pandas DataFrame object
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
20 The target dataset to split
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
21 infile_labels : str
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
22 File path to dataset containing target values
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
23 infile_groups : str
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
24 File path to dataset containing group values
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
25 """
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
26 y = None
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
27 groups = None
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
28
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
29 nth_split = params["mode_selection"]["nth_split"]
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
30
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
31 # read groups
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
32 if infile_groups:
29
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
33 header = (
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
34 "infer"
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
35 if (params["mode_selection"]["cv_selector"]["groups_selector"]["header_g"])
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
36 else None
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
37 )
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
38 column_option = params["mode_selection"]["cv_selector"]["groups_selector"][
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
39 "column_selector_options_g"
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
40 ]["selected_column_selector_option_g"]
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
41 if column_option in [
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
42 "by_index_number",
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
43 "all_but_by_index_number",
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
44 "by_header_name",
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
45 "all_but_by_header_name",
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
46 ]:
29
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
47 c = params["mode_selection"]["cv_selector"]["groups_selector"][
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
48 "column_selector_options_g"
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
49 ]["col_g"]
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
50 else:
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
51 c = None
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
52
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
53 groups = read_columns(
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
54 infile_groups,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
55 c=c,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
56 c_option=column_option,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
57 sep="\t",
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
58 header=header,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
59 parse_dates=True,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
60 )
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
61 groups = groups.ravel()
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
62
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
63 params["mode_selection"]["cv_selector"]["groups_selector"] = groups
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
64
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
65 # read labels
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
66 if infile_labels:
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
67 target_input = params["mode_selection"]["cv_selector"].pop("target_input")
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
68 header = "infer" if target_input["header1"] else None
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
69 col_index = target_input["col"][0] - 1
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
70 df = pd.read_csv(infile_labels, sep="\t", header=header, parse_dates=True)
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
71 y = df.iloc[:, col_index].values
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
72
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
73 # construct the cv splitter object
34
b9d86fc6359d planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 29
diff changeset
74 cv_selector = params["mode_selection"]["cv_selector"]
b9d86fc6359d planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 29
diff changeset
75 if Version(galaxy_ml_version) < Version("0.8.3"):
b9d86fc6359d planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 29
diff changeset
76 cv_selector.pop("n_stratification_bins", None)
b9d86fc6359d planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 29
diff changeset
77 splitter, groups = get_cv(cv_selector)
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
78
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
79 total_n_splits = splitter.get_n_splits(array.values, y=y, groups=groups)
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
80 if nth_split > total_n_splits:
29
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
81 raise ValueError(
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
82 "Total number of splits is {}, but got `nth_split` "
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
83 "= {}".format(total_n_splits, nth_split)
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
84 )
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
85
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
86 i = 1
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
87 for train_index, test_index in splitter.split(array.values, y=y, groups=groups):
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
88 # suppose nth_split >= 1
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
89 if i == nth_split:
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
90 break
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
91 else:
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
92 i += 1
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
93
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
94 train = array.iloc[train_index, :]
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
95 test = array.iloc[test_index, :]
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
96
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
97 return train, test
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
98
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
99
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
100 def main(
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
101 inputs,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
102 infile_array,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
103 outfile_train,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
104 outfile_test,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
105 infile_labels=None,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
106 infile_groups=None,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
107 ):
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
108 """
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
109 Parameter
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
110 ---------
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
111 inputs : str
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
112 File path to galaxy tool parameter
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
113
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
114 infile_array : str
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
115 File paths of input arrays separated by comma
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
116
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
117 infile_labels : str
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
118 File path to dataset containing labels
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
119
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
120 infile_groups : str
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
121 File path to dataset containing groups
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
122
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
123 outfile_train : str
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
124 File path to dataset containing train split
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
125
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
126 outfile_test : str
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
127 File path to dataset containing test split
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
128 """
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
129 warnings.simplefilter("ignore")
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
130
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
131 with open(inputs, "r") as param_handler:
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
132 params = json.load(param_handler)
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
133
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
134 input_header = params["header0"]
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
135 header = "infer" if input_header else None
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
136 array = pd.read_csv(infile_array, sep="\t", header=header, parse_dates=True)
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
137
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
138 # train test split
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
139 if params["mode_selection"]["selected_mode"] == "train_test_split":
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
140 options = params["mode_selection"]["options"]
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
141 shuffle_selection = options.pop("shuffle_selection")
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
142 options["shuffle"] = shuffle_selection["shuffle"]
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
143 if infile_labels:
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
144 header = "infer" if shuffle_selection["header1"] else None
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
145 col_index = shuffle_selection["col"][0] - 1
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
146 df = pd.read_csv(infile_labels, sep="\t", header=header, parse_dates=True)
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
147 labels = df.iloc[:, col_index].values
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
148 options["labels"] = labels
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
149
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
150 train, test = train_test_split(array, **options)
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
151
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
152 # cv splitter
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
153 else:
29
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
154 train, test = _get_single_cv_split(
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
155 params, array, infile_labels=infile_labels, infile_groups=infile_groups
75e9f6fb6536 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents: 28
diff changeset
156 )
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
157
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
158 print("Input shape: %s" % repr(array.shape))
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
159 print("Train shape: %s" % repr(train.shape))
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
160 print("Test shape: %s" % repr(test.shape))
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
161 train.to_csv(outfile_train, sep="\t", header=input_header, index=False)
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
162 test.to_csv(outfile_test, sep="\t", header=input_header, index=False)
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
163
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
164
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
165 if __name__ == "__main__":
23
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
166 aparser = argparse.ArgumentParser()
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
167 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
168 aparser.add_argument("-X", "--infile_array", dest="infile_array")
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
169 aparser.add_argument("-y", "--infile_labels", dest="infile_labels")
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
170 aparser.add_argument("-g", "--infile_groups", dest="infile_groups")
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
171 aparser.add_argument("-o", "--outfile_train", dest="outfile_train")
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
172 aparser.add_argument("-t", "--outfile_test", dest="outfile_test")
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
173 args = aparser.parse_args()
1815eb84736b "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
174
28
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
175 main(
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
176 args.inputs,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
177 args.infile_array,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
178 args.outfile_train,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
179 args.outfile_test,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
180 args.infile_labels,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
181 args.infile_groups,
2ac77e0aec82 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 23
diff changeset
182 )