Mercurial > repos > bgruening > sklearn_feature_selection
annotate stacking_ensembles.py @ 32:a7c667ff83fe draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d970dae980fbe349414fc0889f719d875d999c5b"
author | bgruening |
---|---|
date | Fri, 27 Aug 2021 10:19:00 +0000 |
parents | 5773e98921fc |
children | 61edd9e5c17f |
rev | line source |
---|---|
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
1 import argparse |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
2 import ast |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
3 import json |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
4 import pickle |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
5 import sys |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
6 import warnings |
30
1d20e0dce176
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
29
diff
changeset
|
7 |
1d20e0dce176
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
29
diff
changeset
|
8 import mlxtend.classifier |
1d20e0dce176
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
29
diff
changeset
|
9 import mlxtend.regressor |
1d20e0dce176
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
29
diff
changeset
|
10 import pandas as pd |
31
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
11 from galaxy_ml.utils import (get_cv, get_estimator, get_search_params, |
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
12 load_model) |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
13 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
14 warnings.filterwarnings("ignore") |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
15 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
16 N_JOBS = int(__import__("os").environ.get("GALAXY_SLOTS", 1)) |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
17 |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
18 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
19 def main(inputs_path, output_obj, base_paths=None, meta_path=None, outfile_params=None): |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
20 """ |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
21 Parameter |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
22 --------- |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
23 inputs_path : str |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
24 File path for Galaxy parameters |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
25 |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
26 output_obj : str |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
27 File path for ensemble estimator ouput |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
28 |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
29 base_paths : str |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
30 File path or paths concatenated by comma. |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
31 |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
32 meta_path : str |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
33 File path |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
34 |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
35 outfile_params : str |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
36 File path for params output |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
37 """ |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
38 with open(inputs_path, "r") as param_handler: |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
39 params = json.load(param_handler) |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
40 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
41 estimator_type = params["algo_selection"]["estimator_type"] |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
42 # get base estimators |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
43 base_estimators = [] |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
44 for idx, base_file in enumerate(base_paths.split(",")): |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
45 if base_file and base_file != "None": |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
46 with open(base_file, "rb") as handler: |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
47 model = load_model(handler) |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
48 else: |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
49 estimator_json = params["base_est_builder"][idx]["estimator_selector"] |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
50 model = get_estimator(estimator_json) |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
51 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
52 if estimator_type.startswith("sklearn"): |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
53 named = model.__class__.__name__.lower() |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
54 named = "base_%d_%s" % (idx, named) |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
55 base_estimators.append((named, model)) |
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
56 else: |
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
57 base_estimators.append(model) |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
58 |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
59 # get meta estimator, if applicable |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
60 if estimator_type.startswith("mlxtend"): |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
61 if meta_path: |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
62 with open(meta_path, "rb") as f: |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
63 meta_estimator = load_model(f) |
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
64 else: |
31
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
65 estimator_json = params["algo_selection"]["meta_estimator"][ |
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
66 "estimator_selector" |
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
67 ] |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
68 meta_estimator = get_estimator(estimator_json) |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
69 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
70 options = params["algo_selection"]["options"] |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
71 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
72 cv_selector = options.pop("cv_selector", None) |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
73 if cv_selector: |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
74 splitter, _groups = get_cv(cv_selector) |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
75 options["cv"] = splitter |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
76 # set n_jobs |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
77 options["n_jobs"] = N_JOBS |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
78 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
79 weights = options.pop("weights", None) |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
80 if weights: |
23
8a307b946c58
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
22
diff
changeset
|
81 weights = ast.literal_eval(weights) |
8a307b946c58
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
22
diff
changeset
|
82 if weights: |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
83 options["weights"] = weights |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
84 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
85 mod_and_name = estimator_type.split("_") |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
86 mod = sys.modules[mod_and_name[0]] |
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
87 klass = getattr(mod, mod_and_name[1]) |
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
88 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
89 if estimator_type.startswith("sklearn"): |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
90 options["n_jobs"] = N_JOBS |
20
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
91 ensemble_estimator = klass(base_estimators, **options) |
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
92 |
0b88494bdcac
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
18
diff
changeset
|
93 elif mod == mlxtend.classifier: |
31
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
94 ensemble_estimator = klass( |
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
95 classifiers=base_estimators, meta_classifier=meta_estimator, **options |
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
96 ) |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
97 |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
98 else: |
31
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
99 ensemble_estimator = klass( |
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
100 regressors=base_estimators, meta_regressor=meta_estimator, **options |
5773e98921fc
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents:
30
diff
changeset
|
101 ) |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
102 |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
103 print(ensemble_estimator) |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
104 for base_est in base_estimators: |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
105 print(base_est) |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
106 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
107 with open(output_obj, "wb") as out_handler: |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
108 pickle.dump(ensemble_estimator, out_handler, pickle.HIGHEST_PROTOCOL) |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
109 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
110 if params["get_params"] and outfile_params: |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
111 results = get_search_params(ensemble_estimator) |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
112 df = pd.DataFrame(results, columns=["", "Parameter", "Value"]) |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
113 df.to_csv(outfile_params, sep="\t", index=False) |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
114 |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
115 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
116 if __name__ == "__main__": |
18
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
117 aparser = argparse.ArgumentParser() |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
118 aparser.add_argument("-b", "--bases", dest="bases") |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
119 aparser.add_argument("-m", "--meta", dest="meta") |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
120 aparser.add_argument("-i", "--inputs", dest="inputs") |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
121 aparser.add_argument("-o", "--outfile", dest="outfile") |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
122 aparser.add_argument("-p", "--outfile_params", dest="outfile_params") |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
123 args = aparser.parse_args() |
ec25331946b8
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7
bgruening
parents:
diff
changeset
|
124 |
29
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
125 main( |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
126 args.inputs, |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
127 args.outfile, |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
128 base_paths=args.bases, |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
129 meta_path=args.meta, |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
130 outfile_params=args.outfile_params, |
93f3b307485f
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
23
diff
changeset
|
131 ) |