Mercurial > repos > bgruening > sklearn_clf_metrics
annotate simple_model_fit.py @ 36:40e31fb88836 draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
| author | bgruening | 
|---|---|
| date | Tue, 13 Apr 2021 22:20:07 +0000 | 
| parents | 388f6ec0f3b3 | 
| children | c8547d3993c0 | 
| rev | line source | 
|---|---|
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
1 import argparse | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
2 import json | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
3 import pickle | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
4 | 
| 
36
 
40e31fb88836
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
35 
diff
changeset
 | 
5 import pandas as pd | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
6 from galaxy_ml.utils import load_model, read_columns | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
7 from scipy.io import mmread | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
8 from sklearn.pipeline import Pipeline | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
9 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
10 | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
11 N_JOBS = int(__import__("os").environ.get("GALAXY_SLOTS", 1)) | 
| 
30
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
12 | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
13 | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
14 # TODO import from galaxy_ml.utils in future versions | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
15 def clean_params(estimator, n_jobs=None): | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
16 """clean unwanted hyperparameter settings | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
17 | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
18 If n_jobs is not None, set it into the estimator, if applicable | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
19 | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
20 Return | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
21 ------ | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
22 Cleaned estimator object | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
23 """ | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
24 ALLOWED_CALLBACKS = ( | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
25 "EarlyStopping", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
26 "TerminateOnNaN", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
27 "ReduceLROnPlateau", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
28 "CSVLogger", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
29 "None", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
30 ) | 
| 
30
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
31 | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
32 estimator_params = estimator.get_params() | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
33 | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
34 for name, p in estimator_params.items(): | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
35 # all potential unauthorized file write | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
36 if name == "memory" or name.endswith("__memory") or name.endswith("_path"): | 
| 
30
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
37 new_p = {name: None} | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
38 estimator.set_params(**new_p) | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
39 elif n_jobs is not None and (name == 'n_jobs' or name.endswith('__n_jobs')): | 
| 
30
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
40 new_p = {name: n_jobs} | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
41 estimator.set_params(**new_p) | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
42 elif name.endswith("callbacks"): | 
| 
30
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
43 for cb in p: | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
44 cb_type = cb["callback_selection"]["callback_type"] | 
| 
30
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
45 if cb_type not in ALLOWED_CALLBACKS: | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
46 raise ValueError("Prohibited callback type: %s!" % cb_type) | 
| 
30
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
47 | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
48 return estimator | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
49 | 
| 
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
50 | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
51 def _get_X_y(params, infile1, infile2): | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
52 """read from inputs and output X and y | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
53 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
54 Parameters | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
55 ---------- | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
56 params : dict | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
57 Tool inputs parameter | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
58 infile1 : str | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
59 File path to dataset containing features | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
60 infile2 : str | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
61 File path to dataset containing target values | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
62 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
63 """ | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
64 # store read dataframe object | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
65 loaded_df = {} | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
66 | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
67 input_type = params["input_options"]["selected_input"] | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
68 # tabular input | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
69 if input_type == "tabular": | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
70 header = "infer" if params["input_options"]["header1"] else None | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
71 column_option = params["input_options"]["column_selector_options_1"]["selected_column_selector_option"] | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
72 if column_option in [ | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
73 "by_index_number", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
74 "all_but_by_index_number", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
75 "by_header_name", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
76 "all_but_by_header_name", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
77 ]: | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
78 c = params["input_options"]["column_selector_options_1"]["col1"] | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
79 else: | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
80 c = None | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
81 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
82 df_key = infile1 + repr(header) | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
83 df = pd.read_csv(infile1, sep="\t", header=header, parse_dates=True) | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
84 loaded_df[df_key] = df | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
85 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
86 X = read_columns(df, c=c, c_option=column_option).astype(float) | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
87 # sparse input | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
88 elif input_type == "sparse": | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
89 X = mmread(open(infile1, "r")) | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
90 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
91 # Get target y | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
92 header = "infer" if params["input_options"]["header2"] else None | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
93 column_option = params["input_options"]["column_selector_options_2"]["selected_column_selector_option2"] | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
94 if column_option in [ | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
95 "by_index_number", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
96 "all_but_by_index_number", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
97 "by_header_name", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
98 "all_but_by_header_name", | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
99 ]: | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
100 c = params["input_options"]["column_selector_options_2"]["col2"] | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
101 else: | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
102 c = None | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
103 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
104 df_key = infile2 + repr(header) | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
105 if df_key in loaded_df: | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
106 infile2 = loaded_df[df_key] | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
107 else: | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
108 infile2 = pd.read_csv(infile2, sep="\t", header=header, parse_dates=True) | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
109 loaded_df[df_key] = infile2 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
110 | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
111 y = read_columns(infile2, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
112 c=c, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
113 c_option=column_option, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
114 sep='\t', | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
115 header=header, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
116 parse_dates=True) | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
117 if len(y.shape) == 2 and y.shape[1] == 1: | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
118 y = y.ravel() | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
119 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
120 return X, y | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
121 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
122 | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
123 def main(inputs, infile_estimator, infile1, infile2, out_object, out_weights=None): | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
124 """main | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
125 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
126 Parameters | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
127 ---------- | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
128 inputs : str | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
129 File path to galaxy tool parameter | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
130 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
131 infile_estimator : str | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
132 File paths of input estimator | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
133 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
134 infile1 : str | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
135 File path to dataset containing features | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
136 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
137 infile2 : str | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
138 File path to dataset containing target labels | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
139 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
140 out_object : str | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
141 File path for output of fitted model or skeleton | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
142 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
143 out_weights : str | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
144 File path for output of weights | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
145 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
146 """ | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
147 with open(inputs, "r") as param_handler: | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
148 params = json.load(param_handler) | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
149 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
150 # load model | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
151 with open(infile_estimator, "rb") as est_handler: | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
152 estimator = load_model(est_handler) | 
| 
30
 
9728636c1e41
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
 
bgruening 
parents: 
29 
diff
changeset
 | 
153 estimator = clean_params(estimator, n_jobs=N_JOBS) | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
154 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
155 X_train, y_train = _get_X_y(params, infile1, infile2) | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
156 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
157 estimator.fit(X_train, y_train) | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
158 | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
159 main_est = estimator | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
160 if isinstance(main_est, Pipeline): | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
161 main_est = main_est.steps[-1][-1] | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
162 if hasattr(main_est, "model_") and hasattr(main_est, "save_weights"): | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
163 if out_weights: | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
164 main_est.save_weights(out_weights) | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
165 del main_est.model_ | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
166 del main_est.fit_params | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
167 del main_est.model_class_ | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
168 if getattr(main_est, "validation_data", None): | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
169 del main_est.validation_data | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
170 if getattr(main_est, "data_generator_", None): | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
171 del main_est.data_generator_ | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
172 | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
173 with open(out_object, "wb") as output_handler: | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
174 pickle.dump(estimator, output_handler, pickle.HIGHEST_PROTOCOL) | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
175 | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
176 | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
177 if __name__ == "__main__": | 
| 
29
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
178 aparser = argparse.ArgumentParser() | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
179 aparser.add_argument("-i", "--inputs", dest="inputs", required=True) | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
180 aparser.add_argument("-X", "--infile_estimator", dest="infile_estimator") | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
181 aparser.add_argument("-y", "--infile1", dest="infile1") | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
182 aparser.add_argument("-g", "--infile2", dest="infile2") | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
183 aparser.add_argument("-o", "--out_object", dest="out_object") | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
184 aparser.add_argument("-t", "--out_weights", dest="out_weights") | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
185 args = aparser.parse_args() | 
| 
 
e8ded63c3fa8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
 
bgruening 
parents:  
diff
changeset
 | 
186 | 
| 
35
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
187 main( | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
188 args.inputs, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
189 args.infile_estimator, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
190 args.infile1, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
191 args.infile2, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
192 args.out_object, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
193 args.out_weights, | 
| 
 
388f6ec0f3b3
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
 
bgruening 
parents: 
30 
diff
changeset
 | 
194 ) | 
