Mercurial > repos > bgruening > sklearn_sample_generator

annotate simple_model_fit.py @ 36:8720b49b70da draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
author bgruening
date Mon, 16 Dec 2019 09:43:05 +0000
parents 2d81cd453d14
children 377ddf617ced
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
34
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
1 import argparse
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
2 import json
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
3 import pandas as pd
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
4 import pickle
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
5
d85ad454f15d "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
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
7 from sklearn.pipeline import Pipeline
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
8
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
9
35
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
10 N_JOBS = int(__import__('os').environ.get('GALAXY_SLOTS', 1))
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
11
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
12
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
13 # TODO import from galaxy_ml.utils in future versions
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
14 def clean_params(estimator, n_jobs=None):
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
15 """clean unwanted hyperparameter settings
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
16
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
17 If n_jobs is not None, set it into the estimator, if applicable
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
18
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
19 Return
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
20 ------
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
21 Cleaned estimator object
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
22 """
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
23 ALLOWED_CALLBACKS = ('EarlyStopping', 'TerminateOnNaN',
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
24 'ReduceLROnPlateau', 'CSVLogger', 'None')
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
25
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
26 estimator_params = estimator.get_params()
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
27
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
28 for name, p in estimator_params.items():
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
29 # all potential unauthorized file write
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
30 if name == 'memory' or name.endswith('__memory') \
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
31 or name.endswith('_path'):
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
32 new_p = {name: None}
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
33 estimator.set_params(**new_p)
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
34 elif n_jobs is not None and (name == 'n_jobs' or
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
35 name.endswith('__n_jobs')):
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
36 new_p = {name: n_jobs}
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
37 estimator.set_params(**new_p)
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
38 elif name.endswith('callbacks'):
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
39 for cb in p:
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
40 cb_type = cb['callback_selection']['callback_type']
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
41 if cb_type not in ALLOWED_CALLBACKS:
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
42 raise ValueError(
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
43 "Prohibited callback type: %s!" % cb_type)
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
44
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
45 return estimator
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
46
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
47
34
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
48 def _get_X_y(params, infile1, infile2):
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
49 """ read from inputs and output X and y
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
50
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
51 Parameters
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
52 ----------
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
53 params : dict
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
54 Tool inputs parameter
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
55 infile1 : str
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
56 File path to dataset containing features
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
57 infile2 : str
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
58 File path to dataset containing target values
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
59
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
60 """
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
61 # store read dataframe object
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
62 loaded_df = {}
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
63
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
64 input_type = params['input_options']['selected_input']
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
65 # tabular input
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
66 if input_type == 'tabular':
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
67 header = 'infer' if params['input_options']['header1'] else None
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
68 column_option = (params['input_options']['column_selector_options_1']
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
69 ['selected_column_selector_option'])
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
70 if column_option in ['by_index_number', 'all_but_by_index_number',
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
71 'by_header_name', 'all_but_by_header_name']:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
72 c = params['input_options']['column_selector_options_1']['col1']
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
73 else:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
74 c = None
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
75
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
76 df_key = infile1 + repr(header)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
77 df = pd.read_csv(infile1, sep='\t', header=header,
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
78 parse_dates=True)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
79 loaded_df[df_key] = df
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
80
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
81 X = read_columns(df, c=c, c_option=column_option).astype(float)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
82 # sparse input
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
83 elif input_type == 'sparse':
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
84 X = mmread(open(infile1, 'r'))
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
85
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
86 # Get target y
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
87 header = 'infer' if params['input_options']['header2'] else None
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
88 column_option = (params['input_options']['column_selector_options_2']
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
89 ['selected_column_selector_option2'])
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
90 if column_option in ['by_index_number', 'all_but_by_index_number',
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
91 'by_header_name', 'all_but_by_header_name']:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
92 c = params['input_options']['column_selector_options_2']['col2']
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
93 else:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
94 c = None
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
95
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
96 df_key = infile2 + repr(header)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
97 if df_key in loaded_df:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
98 infile2 = loaded_df[df_key]
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
99 else:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
100 infile2 = pd.read_csv(infile2, sep='\t',
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
101 header=header, parse_dates=True)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
102 loaded_df[df_key] = infile2
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
103
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
104 y = read_columns(
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
105 infile2,
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
106 c=c,
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
107 c_option=column_option,
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
108 sep='\t',
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
109 header=header,
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
110 parse_dates=True)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
111 if len(y.shape) == 2 and y.shape[1] == 1:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
112 y = y.ravel()
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
113
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
114 return X, y
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
115
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
116
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
117 def main(inputs, infile_estimator, infile1, infile2, out_object,
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
118 out_weights=None):
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
119 """ main
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
120
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
121 Parameters
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
122 ----------
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
123 inputs : str
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
124 File path to galaxy tool parameter
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
125
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
126 infile_estimator : str
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
127 File paths of input estimator
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
128
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
129 infile1 : str
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
130 File path to dataset containing features
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
131
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
132 infile2 : str
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
133 File path to dataset containing target labels
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
134
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
135 out_object : str
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
136 File path for output of fitted model or skeleton
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
137
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
138 out_weights : str
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
139 File path for output of weights
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
140
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
141 """
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
142 with open(inputs, 'r') as param_handler:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
143 params = json.load(param_handler)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
144
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
145 # load model
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
146 with open(infile_estimator, 'rb') as est_handler:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
147 estimator = load_model(est_handler)
35
2d81cd453d14 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d6333e7294e67be5968a41f404b66699cad4ae53"
bgruening
parents: 34
diff changeset
148 estimator = clean_params(estimator, n_jobs=N_JOBS)
34
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
149
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
150 X_train, y_train = _get_X_y(params, infile1, infile2)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
151
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
152 estimator.fit(X_train, y_train)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
153
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
154 main_est = estimator
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
155 if isinstance(main_est, Pipeline):
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
156 main_est = main_est.steps[-1][-1]
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
157 if hasattr(main_est, 'model_') \
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
158 and hasattr(main_est, 'save_weights'):
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
159 if out_weights:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
160 main_est.save_weights(out_weights)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
161 del main_est.model_
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
162 del main_est.fit_params
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
163 del main_est.model_class_
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
164 del main_est.validation_data
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
165 if getattr(main_est, 'data_generator_', None):
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
166 del main_est.data_generator_
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
167
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
168 with open(out_object, 'wb') as output_handler:
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
169 pickle.dump(estimator, output_handler,
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
170 pickle.HIGHEST_PROTOCOL)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
171
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
172
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
173 if __name__ == '__main__':
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
174 aparser = argparse.ArgumentParser()
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
175 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
176 aparser.add_argument("-X", "--infile_estimator", dest="infile_estimator")
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
177 aparser.add_argument("-y", "--infile1", dest="infile1")
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
178 aparser.add_argument("-g", "--infile2", dest="infile2")
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
179 aparser.add_argument("-o", "--out_object", dest="out_object")
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
180 aparser.add_argument("-t", "--out_weights", dest="out_weights")
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
181 args = aparser.parse_args()
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
182
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
183 main(args.inputs, args.infile_estimator, args.infile1,
d85ad454f15d "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff changeset
184 args.infile2, args.out_object, args.out_weights)