Mercurial > repos > bgruening > sklearn_stacking_ensemble_models
annotate train_test_split.py @ 11:0380f10c4e04 draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
| author | bgruening | 
|---|---|
| date | Fri, 30 Apr 2021 23:23:56 +0000 | 
| parents | 2d890789ac48 | 
| children | 
| 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 warnings | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
4 | 
| 
10
 
2d890789ac48
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
9 
diff
changeset
 | 
5 import pandas as pd | 
| 
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: 
5 
diff
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: 
5 
diff
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: 
5 
diff
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: | 
| 
11
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
31 header = ( | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
32 "infer" | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
33 if (params["mode_selection"]["cv_selector"]["groups_selector"]["header_g"]) | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
34 else None | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
35 ) | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
36 column_option = params["mode_selection"]["cv_selector"]["groups_selector"][ | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
37 "column_selector_options_g" | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
38 ]["selected_column_selector_option_g"] | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
39 if column_option in [ | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
40 "by_index_number", | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
41 "all_but_by_index_number", | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
42 "by_header_name", | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
43 "all_but_by_header_name", | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
44 ]: | 
| 
11
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
45 c = params["mode_selection"]["cv_selector"]["groups_selector"][ | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
46 "column_selector_options_g" | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
47 ]["col_g"] | 
| 
5
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
48 else: | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
49 c = None | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
50 | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
51 groups = read_columns( | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
52 infile_groups, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
53 c=c, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
54 c_option=column_option, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
55 sep="\t", | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
56 header=header, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
57 parse_dates=True, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
58 ) | 
| 
5
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
59 groups = groups.ravel() | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
60 | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
61 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
 | 
62 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
63 # read labels | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
64 if infile_labels: | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
65 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: 
5 
diff
changeset
 | 
66 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: 
5 
diff
changeset
 | 
67 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: 
5 
diff
changeset
 | 
68 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
 | 
69 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
 | 
70 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
71 # construct the cv splitter object | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
72 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
 | 
73 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
74 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
 | 
75 if nth_split > total_n_splits: | 
| 
11
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
76 raise ValueError( | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
77 "Total number of splits is {}, but got `nth_split` " | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
78 "= {}".format(total_n_splits, nth_split) | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
79 ) | 
| 
5
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
80 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
81 i = 1 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
82 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
 | 
83 # suppose nth_split >= 1 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
84 if i == nth_split: | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
85 break | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
86 else: | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
87 i += 1 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
88 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
89 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
 | 
90 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
 | 
91 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
92 return train, test | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
93 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
94 | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
95 def main( | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
96 inputs, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
97 infile_array, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
98 outfile_train, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
99 outfile_test, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
100 infile_labels=None, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
101 infile_groups=None, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
102 ): | 
| 
5
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
103 """ | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
104 Parameter | 
| 
 
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 inputs : 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 galaxy tool parameter | 
| 
 
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 infile_array : str | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
110 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
 | 
111 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
112 infile_labels : 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 labels | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
114 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
115 infile_groups : str | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
116 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
 | 
117 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
118 outfile_train : str | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
119 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
 | 
120 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
121 outfile_test : str | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
122 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
 | 
123 """ | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
124 warnings.simplefilter("ignore") | 
| 
5
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
125 | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
126 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
 | 
127 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
 | 
128 | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
129 input_header = params["header0"] | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
130 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: 
5 
diff
changeset
 | 
131 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
 | 
132 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
133 # train test split | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
134 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: 
5 
diff
changeset
 | 
135 options = params["mode_selection"]["options"] | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
136 shuffle_selection = options.pop("shuffle_selection") | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
137 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
 | 
138 if infile_labels: | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
139 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: 
5 
diff
changeset
 | 
140 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: 
5 
diff
changeset
 | 
141 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
 | 
142 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: 
5 
diff
changeset
 | 
143 options["labels"] = labels | 
| 
5
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
144 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
145 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
 | 
146 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
147 # cv splitter | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
148 else: | 
| 
11
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
149 train, test = _get_single_cv_split( | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
150 params, array, infile_labels=infile_labels, infile_groups=infile_groups | 
| 
 
0380f10c4e04
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
10 
diff
changeset
 | 
151 ) | 
| 
5
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
152 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
153 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
 | 
154 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
 | 
155 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: 
5 
diff
changeset
 | 
156 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: 
5 
diff
changeset
 | 
157 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
 | 
158 | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
159 | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
160 if __name__ == "__main__": | 
| 
5
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
161 aparser = argparse.ArgumentParser() | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
162 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
 | 
163 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
 | 
164 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
 | 
165 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
 | 
166 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
 | 
167 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
 | 
168 args = aparser.parse_args() | 
| 
 
8b5b653ba1ed
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
169 | 
| 
9
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
170 main( | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
171 args.inputs, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
172 args.infile_array, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
173 args.outfile_train, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
174 args.outfile_test, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
175 args.infile_labels, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
176 args.infile_groups, | 
| 
 
b8c92e94ac1d
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
5 
diff
changeset
 | 
177 ) | 
