Mercurial > repos > bgruening > sklearn_stacking_ensemble_models
annotate train_test_split.py @ 9:b8c92e94ac1d draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
| author | bgruening | 
|---|---|
| date | Tue, 13 Apr 2021 15:49:42 +0000 | 
| parents | 8b5b653ba1ed | 
| children | 2d890789ac48 | 
| rev | line source | 
|---|---|
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 1 import argparse | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 2 import json | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 3 import pandas as pd | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 4 import warnings | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 5 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 6 from galaxy_ml.model_validations import train_test_split | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 7 from galaxy_ml.utils import get_cv, read_columns | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 8 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 9 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 10 def _get_single_cv_split(params, array, infile_labels=None, infile_groups=None): | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 11 """output (train, test) subset from a cv splitter | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 12 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 13 Parameters | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 14 ---------- | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 15 params : dict | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 16 Galaxy tool inputs | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 17 array : pandas DataFrame object | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 18 The target dataset to split | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 19 infile_labels : str | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 20 File path to dataset containing target values | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 21 infile_groups : str | 
| 
8b5b653ba1ed
"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 group values | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 23 """ | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 24 y = None | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 25 groups = None | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 26 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 27 nth_split = params["mode_selection"]["nth_split"] | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 28 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 29 # read groups | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 30 if infile_groups: | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 31 header = "infer" if (params["mode_selection"]["cv_selector"]["groups_selector"]["header_g"]) else None | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 32 column_option = params["mode_selection"]["cv_selector"]["groups_selector"]["column_selector_options_g"][ | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 33 "selected_column_selector_option_g" | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 34 ] | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 35 if column_option in [ | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 36 "by_index_number", | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 37 "all_but_by_index_number", | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 38 "by_header_name", | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 39 "all_but_by_header_name", | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 40 ]: | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 41 c = params["mode_selection"]["cv_selector"]["groups_selector"]["column_selector_options_g"]["col_g"] | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 42 else: | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 43 c = None | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 44 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 45 groups = read_columns( | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 46 infile_groups, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 47 c=c, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 48 c_option=column_option, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 49 sep="\t", | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 50 header=header, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 51 parse_dates=True, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 52 ) | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 53 groups = groups.ravel() | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 54 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 55 params["mode_selection"]["cv_selector"]["groups_selector"] = groups | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 56 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 57 # read labels | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 58 if infile_labels: | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 59 target_input = params["mode_selection"]["cv_selector"].pop("target_input") | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 60 header = "infer" if target_input["header1"] else None | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 61 col_index = target_input["col"][0] - 1 | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 62 df = pd.read_csv(infile_labels, sep="\t", header=header, parse_dates=True) | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 63 y = df.iloc[:, col_index].values | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 64 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 65 # construct the cv splitter object | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 66 splitter, groups = get_cv(params["mode_selection"]["cv_selector"]) | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 67 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 68 total_n_splits = splitter.get_n_splits(array.values, y=y, groups=groups) | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 69 if nth_split > total_n_splits: | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 70 raise ValueError("Total number of splits is {}, but got `nth_split` " "= {}".format(total_n_splits, nth_split)) | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 71 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 72 i = 1 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 73 for train_index, test_index in splitter.split(array.values, y=y, groups=groups): | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 74 # suppose nth_split >= 1 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 75 if i == nth_split: | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 76 break | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 77 else: | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 78 i += 1 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 79 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 80 train = array.iloc[train_index, :] | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 81 test = array.iloc[test_index, :] | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 82 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 83 return train, test | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 84 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 85 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 86 def main( | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 87 inputs, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 88 infile_array, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 89 outfile_train, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 90 outfile_test, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 91 infile_labels=None, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 92 infile_groups=None, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 93 ): | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 94 """ | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 95 Parameter | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 96 --------- | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 97 inputs : str | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 98 File path to galaxy tool parameter | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 99 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 100 infile_array : str | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 101 File paths of input arrays separated by comma | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 102 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 103 infile_labels : str | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 104 File path to dataset containing labels | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 105 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 106 infile_groups : str | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 107 File path to dataset containing groups | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 108 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 109 outfile_train : str | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 110 File path to dataset containing train split | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 111 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 112 outfile_test : str | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 113 File path to dataset containing test split | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 114 """ | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 115 warnings.simplefilter("ignore") | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 116 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 117 with open(inputs, "r") as param_handler: | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 118 params = json.load(param_handler) | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 119 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 120 input_header = params["header0"] | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 121 header = "infer" if input_header else None | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 122 array = pd.read_csv(infile_array, sep="\t", header=header, parse_dates=True) | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 123 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 124 # train test split | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 125 if params["mode_selection"]["selected_mode"] == "train_test_split": | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 126 options = params["mode_selection"]["options"] | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 127 shuffle_selection = options.pop("shuffle_selection") | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 128 options["shuffle"] = shuffle_selection["shuffle"] | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 129 if infile_labels: | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 130 header = "infer" if shuffle_selection["header1"] else None | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 131 col_index = shuffle_selection["col"][0] - 1 | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 132 df = pd.read_csv(infile_labels, sep="\t", header=header, parse_dates=True) | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 133 labels = df.iloc[:, col_index].values | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 134 options["labels"] = labels | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 135 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 136 train, test = train_test_split(array, **options) | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 137 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 138 # cv splitter | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 139 else: | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 140 train, test = _get_single_cv_split(params, array, infile_labels=infile_labels, infile_groups=infile_groups) | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 141 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 142 print("Input shape: %s" % repr(array.shape)) | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 143 print("Train shape: %s" % repr(train.shape)) | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 144 print("Test shape: %s" % repr(test.shape)) | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 145 train.to_csv(outfile_train, sep="\t", header=input_header, index=False) | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 146 test.to_csv(outfile_test, sep="\t", header=input_header, index=False) | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 147 | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 148 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 149 if __name__ == "__main__": | 
| 5 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 150 aparser = argparse.ArgumentParser() | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 151 aparser.add_argument("-i", "--inputs", dest="inputs", required=True) | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 152 aparser.add_argument("-X", "--infile_array", dest="infile_array") | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 153 aparser.add_argument("-y", "--infile_labels", dest="infile_labels") | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 154 aparser.add_argument("-g", "--infile_groups", dest="infile_groups") | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 155 aparser.add_argument("-o", "--outfile_train", dest="outfile_train") | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 156 aparser.add_argument("-t", "--outfile_test", dest="outfile_test") | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 157 args = aparser.parse_args() | 
| 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 bgruening parents: diff
changeset | 158 | 
| 9 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 159 main( | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 160 args.inputs, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 161 args.infile_array, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 162 args.outfile_train, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 163 args.outfile_test, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 164 args.infile_labels, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 165 args.infile_groups, | 
| 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 bgruening parents: 
5diff
changeset | 166 ) | 
