Mercurial > repos > bgruening > sklearn_svm_classifier
annotate stacking_ensembles.py @ 10:153f237ddb36 draft
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
author | bgruening |
---|---|
date | Fri, 09 Aug 2019 07:08:07 -0400 |
parents | 1a9d5a8fff12 |
children | 73d2ef652879 |
rev | line source |
---|---|
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
1 import argparse |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
2 import ast |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
3 import json |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
4 import mlxtend.regressor |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
5 import mlxtend.classifier |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
6 import pandas as pd |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
7 import pickle |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
8 import sklearn |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
9 import sys |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
10 import warnings |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
11 from sklearn import ensemble |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
12 |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
13 from galaxy_ml.utils import (load_model, get_cv, get_estimator, |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
14 get_search_params) |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
15 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
16 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
17 warnings.filterwarnings('ignore') |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
18 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
19 N_JOBS = int(__import__('os').environ.get('GALAXY_SLOTS', 1)) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
20 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
21 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
22 def main(inputs_path, output_obj, base_paths=None, meta_path=None, |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
23 outfile_params=None): |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
24 """ |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
25 Parameter |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
26 --------- |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
27 inputs_path : str |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
28 File path for Galaxy parameters |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
29 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
30 output_obj : str |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
31 File path for ensemble estimator ouput |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
32 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
33 base_paths : str |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
34 File path or paths concatenated by comma. |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
35 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
36 meta_path : str |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
37 File path |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
38 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
39 outfile_params : str |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
40 File path for params output |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
41 """ |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
42 with open(inputs_path, 'r') as param_handler: |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
43 params = json.load(param_handler) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
44 |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
45 estimator_type = params['algo_selection']['estimator_type'] |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
46 # get base estimators |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
47 base_estimators = [] |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
48 for idx, base_file in enumerate(base_paths.split(',')): |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
49 if base_file and base_file != 'None': |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
50 with open(base_file, 'rb') as handler: |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
51 model = load_model(handler) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
52 else: |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
53 estimator_json = (params['base_est_builder'][idx] |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
54 ['estimator_selector']) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
55 model = get_estimator(estimator_json) |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
56 |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
57 if estimator_type.startswith('sklearn'): |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
58 named = model.__class__.__name__.lower() |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
59 named = 'base_%d_%s' % (idx, named) |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
60 base_estimators.append((named, model)) |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
61 else: |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
62 base_estimators.append(model) |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
63 |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
64 # get meta estimator, if applicable |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
65 if estimator_type.startswith('mlxtend'): |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
66 if meta_path: |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
67 with open(meta_path, 'rb') as f: |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
68 meta_estimator = load_model(f) |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
69 else: |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
70 estimator_json = (params['algo_selection'] |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
71 ['meta_estimator']['estimator_selector']) |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
72 meta_estimator = get_estimator(estimator_json) |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
73 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
74 options = params['algo_selection']['options'] |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
75 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
76 cv_selector = options.pop('cv_selector', None) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
77 if cv_selector: |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
78 splitter, groups = get_cv(cv_selector) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
79 options['cv'] = splitter |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
80 # set n_jobs |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
81 options['n_jobs'] = N_JOBS |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
82 |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
83 weights = options.pop('weights', None) |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
84 if weights: |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
85 options['weights'] = ast.literal_eval(weights) |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
86 |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
87 mod_and_name = estimator_type.split('_') |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
88 mod = sys.modules[mod_and_name[0]] |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
89 klass = getattr(mod, mod_and_name[1]) |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
90 |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
91 if estimator_type.startswith('sklearn'): |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
92 options['n_jobs'] = N_JOBS |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
93 ensemble_estimator = klass(base_estimators, **options) |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
94 |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
95 elif mod == mlxtend.classifier: |
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
96 ensemble_estimator = klass( |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
97 classifiers=base_estimators, |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
98 meta_classifier=meta_estimator, |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
99 **options) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
100 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
101 else: |
10
153f237ddb36
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
8
diff
changeset
|
102 ensemble_estimator = klass( |
8
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
103 regressors=base_estimators, |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
104 meta_regressor=meta_estimator, |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
105 **options) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
106 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
107 print(ensemble_estimator) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
108 for base_est in base_estimators: |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
109 print(base_est) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
110 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
111 with open(output_obj, 'wb') as out_handler: |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
112 pickle.dump(ensemble_estimator, out_handler, pickle.HIGHEST_PROTOCOL) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
113 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
114 if params['get_params'] and outfile_params: |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
115 results = get_search_params(ensemble_estimator) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
116 df = pd.DataFrame(results, columns=['', 'Parameter', 'Value']) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
117 df.to_csv(outfile_params, sep='\t', index=False) |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
118 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
119 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
120 if __name__ == '__main__': |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
121 aparser = argparse.ArgumentParser() |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
122 aparser.add_argument("-b", "--bases", dest="bases") |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
123 aparser.add_argument("-m", "--meta", dest="meta") |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
124 aparser.add_argument("-i", "--inputs", dest="inputs") |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
125 aparser.add_argument("-o", "--outfile", dest="outfile") |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
126 aparser.add_argument("-p", "--outfile_params", dest="outfile_params") |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
127 args = aparser.parse_args() |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
128 |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
129 main(args.inputs, args.outfile, base_paths=args.bases, |
1a9d5a8fff12
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
130 meta_path=args.meta, outfile_params=args.outfile_params) |