Mercurial > repos > bgruening > keras_batch_models
annotate stacking_ensembles.py @ 8:8edfc7381344 draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
author | bgruening |
---|---|
date | Tue, 13 Apr 2021 15:58:22 +0000 |
parents | 03d0ca500753 |
children | 5369fdfec6a6 |
rev | line source |
---|---|
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
1 import argparse |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
2 import ast |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
3 import json |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
4 import mlxtend.regressor |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
5 import mlxtend.classifier |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
6 import pandas as pd |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
7 import pickle |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
8 import sys |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
9 import warnings |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
10 from galaxy_ml.utils import load_model, get_cv, get_estimator, get_search_params |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
11 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
12 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
13 warnings.filterwarnings("ignore") |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
14 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
15 N_JOBS = int(__import__("os").environ.get("GALAXY_SLOTS", 1)) |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
16 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
17 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
18 def main(inputs_path, output_obj, base_paths=None, meta_path=None, outfile_params=None): |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
19 """ |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
20 Parameter |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
21 --------- |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
22 inputs_path : str |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
23 File path for Galaxy parameters |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
24 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
25 output_obj : str |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
26 File path for ensemble estimator ouput |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
27 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
28 base_paths : str |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
29 File path or paths concatenated by comma. |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
30 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
31 meta_path : str |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
32 File path |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
33 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
34 outfile_params : str |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
35 File path for params output |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
36 """ |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
37 with open(inputs_path, "r") as param_handler: |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
38 params = json.load(param_handler) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
39 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
40 estimator_type = params["algo_selection"]["estimator_type"] |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
41 # get base estimators |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
42 base_estimators = [] |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
43 for idx, base_file in enumerate(base_paths.split(",")): |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
44 if base_file and base_file != "None": |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
45 with open(base_file, "rb") as handler: |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
46 model = load_model(handler) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
47 else: |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
48 estimator_json = params["base_est_builder"][idx]["estimator_selector"] |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
49 model = get_estimator(estimator_json) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
50 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
51 if estimator_type.startswith("sklearn"): |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
52 named = model.__class__.__name__.lower() |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
53 named = "base_%d_%s" % (idx, named) |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
54 base_estimators.append((named, model)) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
55 else: |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
56 base_estimators.append(model) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
57 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
58 # get meta estimator, if applicable |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
59 if estimator_type.startswith("mlxtend"): |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
60 if meta_path: |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
61 with open(meta_path, "rb") as f: |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
62 meta_estimator = load_model(f) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
63 else: |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
64 estimator_json = params["algo_selection"]["meta_estimator"]["estimator_selector"] |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
65 meta_estimator = get_estimator(estimator_json) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
66 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
67 options = params["algo_selection"]["options"] |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
68 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
69 cv_selector = options.pop("cv_selector", None) |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
70 if cv_selector: |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
71 splitter, _groups = get_cv(cv_selector) |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
72 options["cv"] = splitter |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
73 # set n_jobs |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
74 options["n_jobs"] = N_JOBS |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
75 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
76 weights = options.pop("weights", None) |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
77 if weights: |
3
03d0ca500753
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
2
diff
changeset
|
78 weights = ast.literal_eval(weights) |
03d0ca500753
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
2
diff
changeset
|
79 if weights: |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
80 options["weights"] = weights |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
81 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
82 mod_and_name = estimator_type.split("_") |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
83 mod = sys.modules[mod_and_name[0]] |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
84 klass = getattr(mod, mod_and_name[1]) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
85 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
86 if estimator_type.startswith("sklearn"): |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
87 options["n_jobs"] = N_JOBS |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
88 ensemble_estimator = klass(base_estimators, **options) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
89 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
90 elif mod == mlxtend.classifier: |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
91 ensemble_estimator = klass(classifiers=base_estimators, meta_classifier=meta_estimator, **options) |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
92 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
93 else: |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
94 ensemble_estimator = klass(regressors=base_estimators, meta_regressor=meta_estimator, **options) |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
95 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
96 print(ensemble_estimator) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
97 for base_est in base_estimators: |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
98 print(base_est) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
99 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
100 with open(output_obj, "wb") as out_handler: |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
101 pickle.dump(ensemble_estimator, out_handler, pickle.HIGHEST_PROTOCOL) |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
102 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
103 if params["get_params"] and outfile_params: |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
104 results = get_search_params(ensemble_estimator) |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
105 df = pd.DataFrame(results, columns=["", "Parameter", "Value"]) |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
106 df.to_csv(outfile_params, sep="\t", index=False) |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
107 |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
108 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
109 if __name__ == "__main__": |
0
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
110 aparser = argparse.ArgumentParser() |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
111 aparser.add_argument("-b", "--bases", dest="bases") |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
112 aparser.add_argument("-m", "--meta", dest="meta") |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
113 aparser.add_argument("-i", "--inputs", dest="inputs") |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
114 aparser.add_argument("-o", "--outfile", dest="outfile") |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
115 aparser.add_argument("-p", "--outfile_params", dest="outfile_params") |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
116 args = aparser.parse_args() |
f59a4f7c47f9
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 60f0fbc0eafd7c11bc60fb6c77f2937782efd8a9-dirty
bgruening
parents:
diff
changeset
|
117 |
8
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
118 main( |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
119 args.inputs, |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
120 args.outfile, |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
121 base_paths=args.bases, |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
122 meta_path=args.meta, |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
123 outfile_params=args.outfile_params, |
8edfc7381344
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents:
3
diff
changeset
|
124 ) |