annotate search_model_validation.py @ 1:8ddb3557710d draft

"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
author bgruening
date Wed, 22 Jan 2020 08:00:08 -0500
parents 03f61bb3ca43
children 3866911c93ae
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
1 import argparse
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
2 import collections
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
3 import imblearn
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
4 import joblib
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
5 import json
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
6 import numpy as np
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
7 import os
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
8 import pandas as pd
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
9 import pickle
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
10 import skrebate
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
11 import sys
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
12 import warnings
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
13 from scipy.io import mmread
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
14 from sklearn import (cluster, decomposition, feature_selection,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
15 kernel_approximation, model_selection, preprocessing)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
16 from sklearn.exceptions import FitFailedWarning
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
17 from sklearn.model_selection._validation import _score, cross_validate
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
18 from sklearn.model_selection import _search, _validation
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
19 from sklearn.pipeline import Pipeline
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
20
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
21 from galaxy_ml.utils import (SafeEval, get_cv, get_scoring, load_model,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
22 read_columns, try_get_attr, get_module,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
23 clean_params, get_main_estimator)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
24
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
25
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
26 _fit_and_score = try_get_attr('galaxy_ml.model_validations', '_fit_and_score')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
27 setattr(_search, '_fit_and_score', _fit_and_score)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
28 setattr(_validation, '_fit_and_score', _fit_and_score)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
29
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
30 N_JOBS = int(os.environ.get('GALAXY_SLOTS', 1))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
31 # handle disk cache
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
32 CACHE_DIR = os.path.join(os.getcwd(), 'cached')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
33 del os
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
34 NON_SEARCHABLE = ('n_jobs', 'pre_dispatch', 'memory', '_path',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
35 'nthread', 'callbacks')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
36
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
37
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
38 def _eval_search_params(params_builder):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
39 search_params = {}
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
40
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
41 for p in params_builder['param_set']:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
42 search_list = p['sp_list'].strip()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
43 if search_list == '':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
44 continue
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
45
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
46 param_name = p['sp_name']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
47 if param_name.lower().endswith(NON_SEARCHABLE):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
48 print("Warning: `%s` is not eligible for search and was "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
49 "omitted!" % param_name)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
50 continue
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
51
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
52 if not search_list.startswith(':'):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
53 safe_eval = SafeEval(load_scipy=True, load_numpy=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
54 ev = safe_eval(search_list)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
55 search_params[param_name] = ev
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
56 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
57 # Have `:` before search list, asks for estimator evaluatio
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
58 safe_eval_es = SafeEval(load_estimators=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
59 search_list = search_list[1:].strip()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
60 # TODO maybe add regular express check
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
61 ev = safe_eval_es(search_list)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
62 preprocessings = (
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
63 preprocessing.StandardScaler(), preprocessing.Binarizer(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
64 preprocessing.MaxAbsScaler(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
65 preprocessing.Normalizer(), preprocessing.MinMaxScaler(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
66 preprocessing.PolynomialFeatures(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
67 preprocessing.RobustScaler(), feature_selection.SelectKBest(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
68 feature_selection.GenericUnivariateSelect(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
69 feature_selection.SelectPercentile(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
70 feature_selection.SelectFpr(), feature_selection.SelectFdr(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
71 feature_selection.SelectFwe(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
72 feature_selection.VarianceThreshold(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
73 decomposition.FactorAnalysis(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
74 decomposition.FastICA(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
75 decomposition.IncrementalPCA(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
76 decomposition.KernelPCA(random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
77 decomposition.LatentDirichletAllocation(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
78 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
79 decomposition.MiniBatchDictionaryLearning(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
80 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
81 decomposition.MiniBatchSparsePCA(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
82 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
83 decomposition.NMF(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
84 decomposition.PCA(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
85 decomposition.SparsePCA(random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
86 decomposition.TruncatedSVD(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
87 kernel_approximation.Nystroem(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
88 kernel_approximation.RBFSampler(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
89 kernel_approximation.AdditiveChi2Sampler(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
90 kernel_approximation.SkewedChi2Sampler(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
91 cluster.FeatureAgglomeration(),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
92 skrebate.ReliefF(n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
93 skrebate.SURF(n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
94 skrebate.SURFstar(n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
95 skrebate.MultiSURF(n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
96 skrebate.MultiSURFstar(n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
97 imblearn.under_sampling.ClusterCentroids(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
98 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
99 imblearn.under_sampling.CondensedNearestNeighbour(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
100 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
101 imblearn.under_sampling.EditedNearestNeighbours(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
102 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
103 imblearn.under_sampling.RepeatedEditedNearestNeighbours(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
104 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
105 imblearn.under_sampling.AllKNN(random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
106 imblearn.under_sampling.InstanceHardnessThreshold(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
107 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
108 imblearn.under_sampling.NearMiss(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
109 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
110 imblearn.under_sampling.NeighbourhoodCleaningRule(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
111 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
112 imblearn.under_sampling.OneSidedSelection(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
113 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
114 imblearn.under_sampling.RandomUnderSampler(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
115 random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
116 imblearn.under_sampling.TomekLinks(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
117 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
118 imblearn.over_sampling.ADASYN(random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
119 imblearn.over_sampling.RandomOverSampler(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
120 imblearn.over_sampling.SMOTE(random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
121 imblearn.over_sampling.SVMSMOTE(random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
122 imblearn.over_sampling.BorderlineSMOTE(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
123 random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
124 imblearn.over_sampling.SMOTENC(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
125 categorical_features=[], random_state=0, n_jobs=N_JOBS),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
126 imblearn.combine.SMOTEENN(random_state=0),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
127 imblearn.combine.SMOTETomek(random_state=0))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
128 newlist = []
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
129 for obj in ev:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
130 if obj is None:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
131 newlist.append(None)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
132 elif obj == 'all_0':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
133 newlist.extend(preprocessings[0:35])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
134 elif obj == 'sk_prep_all': # no KernalCenter()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
135 newlist.extend(preprocessings[0:7])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
136 elif obj == 'fs_all':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
137 newlist.extend(preprocessings[7:14])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
138 elif obj == 'decomp_all':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
139 newlist.extend(preprocessings[14:25])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
140 elif obj == 'k_appr_all':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
141 newlist.extend(preprocessings[25:29])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
142 elif obj == 'reb_all':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
143 newlist.extend(preprocessings[30:35])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
144 elif obj == 'imb_all':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
145 newlist.extend(preprocessings[35:54])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
146 elif type(obj) is int and -1 < obj < len(preprocessings):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
147 newlist.append(preprocessings[obj])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
148 elif hasattr(obj, 'get_params'): # user uploaded object
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
149 if 'n_jobs' in obj.get_params():
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
150 newlist.append(obj.set_params(n_jobs=N_JOBS))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
151 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
152 newlist.append(obj)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
153 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
154 sys.exit("Unsupported estimator type: %r" % (obj))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
155
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
156 search_params[param_name] = newlist
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
157
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
158 return search_params
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
159
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
160
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
161 def _handle_X_y(estimator, params, infile1, infile2, loaded_df={},
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
162 ref_seq=None, intervals=None, targets=None,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
163 fasta_path=None):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
164 """read inputs
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
165
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
166 Params
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
167 -------
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
168 estimator : estimator object
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
169 params : dict
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
170 Galaxy tool parameter inputs
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
171 infile1 : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
172 File path to dataset containing features
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
173 infile2 : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
174 File path to dataset containing target values
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
175 loaded_df : dict
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
176 Contains loaded DataFrame objects with file path as keys
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
177 ref_seq : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
178 File path to dataset containing genome sequence file
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
179 interval : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
180 File path to dataset containing interval file
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
181 targets : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
182 File path to dataset compressed target bed file
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
183 fasta_path : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
184 File path to dataset containing fasta file
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
185
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
186
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
187 Returns
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
188 -------
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
189 estimator : estimator object after setting new attributes
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
190 X : numpy array
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
191 y : numpy array
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
192 """
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
193 estimator_params = estimator.get_params()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
194
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
195 input_type = params['input_options']['selected_input']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
196 # tabular input
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
197 if input_type == 'tabular':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
198 header = 'infer' if params['input_options']['header1'] else None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
199 column_option = (params['input_options']['column_selector_options_1']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
200 ['selected_column_selector_option'])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
201 if column_option in ['by_index_number', 'all_but_by_index_number',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
202 'by_header_name', 'all_but_by_header_name']:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
203 c = params['input_options']['column_selector_options_1']['col1']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
204 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
205 c = None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
206
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
207 df_key = infile1 + repr(header)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
208
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
209 if df_key in loaded_df:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
210 infile1 = loaded_df[df_key]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
211
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
212 df = pd.read_csv(infile1, sep='\t', header=header,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
213 parse_dates=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
214 loaded_df[df_key] = df
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
215
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
216 X = read_columns(df, c=c, c_option=column_option).astype(float)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
217 # sparse input
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
218 elif input_type == 'sparse':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
219 X = mmread(open(infile1, 'r'))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
220
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
221 # fasta_file input
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
222 elif input_type == 'seq_fasta':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
223 pyfaidx = get_module('pyfaidx')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
224 sequences = pyfaidx.Fasta(fasta_path)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
225 n_seqs = len(sequences.keys())
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
226 X = np.arange(n_seqs)[:, np.newaxis]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
227 for param in estimator_params.keys():
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
228 if param.endswith('fasta_path'):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
229 estimator.set_params(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
230 **{param: fasta_path})
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
231 break
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
232 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
233 raise ValueError(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
234 "The selected estimator doesn't support "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
235 "fasta file input! Please consider using "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
236 "KerasGBatchClassifier with "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
237 "FastaDNABatchGenerator/FastaProteinBatchGenerator "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
238 "or having GenomeOneHotEncoder/ProteinOneHotEncoder "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
239 "in pipeline!")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
240
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
241 elif input_type == 'refseq_and_interval':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
242 path_params = {
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
243 'data_batch_generator__ref_genome_path': ref_seq,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
244 'data_batch_generator__intervals_path': intervals,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
245 'data_batch_generator__target_path': targets
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
246 }
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
247 estimator.set_params(**path_params)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
248 n_intervals = sum(1 for line in open(intervals))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
249 X = np.arange(n_intervals)[:, np.newaxis]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
250
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
251 # Get target y
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
252 header = 'infer' if params['input_options']['header2'] else None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
253 column_option = (params['input_options']['column_selector_options_2']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
254 ['selected_column_selector_option2'])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
255 if column_option in ['by_index_number', 'all_but_by_index_number',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
256 'by_header_name', 'all_but_by_header_name']:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
257 c = params['input_options']['column_selector_options_2']['col2']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
258 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
259 c = None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
260
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
261 df_key = infile2 + repr(header)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
262 if df_key in loaded_df:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
263 infile2 = loaded_df[df_key]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
264 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
265 infile2 = pd.read_csv(infile2, sep='\t',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
266 header=header, parse_dates=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
267 loaded_df[df_key] = infile2
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
268
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
269 y = read_columns(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
270 infile2,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
271 c=c,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
272 c_option=column_option,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
273 sep='\t',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
274 header=header,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
275 parse_dates=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
276 if len(y.shape) == 2 and y.shape[1] == 1:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
277 y = y.ravel()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
278 if input_type == 'refseq_and_interval':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
279 estimator.set_params(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
280 data_batch_generator__features=y.ravel().tolist())
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
281 y = None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
282 # end y
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
283
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
284 return estimator, X, y
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
285
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
286
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
287 def _do_outer_cv(searcher, X, y, outer_cv, scoring, error_score='raise',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
288 outfile=None):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
289 """Do outer cross-validation for nested CV
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
290
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
291 Parameters
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
292 ----------
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
293 searcher : object
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
294 SearchCV object
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
295 X : numpy array
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
296 Containing features
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
297 y : numpy array
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
298 Target values or labels
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
299 outer_cv : int or CV splitter
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
300 Control the cv splitting
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
301 scoring : object
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
302 Scorer
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
303 error_score: str, float or numpy float
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
304 Whether to raise fit error or return an value
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
305 outfile : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
306 File path to store the restuls
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
307 """
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
308 if error_score == 'raise':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
309 rval = cross_validate(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
310 searcher, X, y, scoring=scoring,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
311 cv=outer_cv, n_jobs=N_JOBS, verbose=0,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
312 error_score=error_score)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
313 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
314 warnings.simplefilter('always', FitFailedWarning)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
315 with warnings.catch_warnings(record=True) as w:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
316 try:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
317 rval = cross_validate(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
318 searcher, X, y,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
319 scoring=scoring,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
320 cv=outer_cv, n_jobs=N_JOBS,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
321 verbose=0,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
322 error_score=error_score)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
323 except ValueError:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
324 pass
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
325 for warning in w:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
326 print(repr(warning.message))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
327
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
328 keys = list(rval.keys())
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
329 for k in keys:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
330 if k.startswith('test'):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
331 rval['mean_' + k] = np.mean(rval[k])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
332 rval['std_' + k] = np.std(rval[k])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
333 if k.endswith('time'):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
334 rval.pop(k)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
335 rval = pd.DataFrame(rval)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
336 rval = rval[sorted(rval.columns)]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
337 rval.to_csv(path_or_buf=outfile, sep='\t', header=True, index=False)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
338
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
339
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
340 def _do_train_test_split_val(searcher, X, y, params, error_score='raise',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
341 primary_scoring=None, groups=None,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
342 outfile=None):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
343 """ do train test split, searchCV validates on the train and then use
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
344 the best_estimator_ to evaluate on the test
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
345
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
346 Returns
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
347 --------
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
348 Fitted SearchCV object
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
349 """
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
350 train_test_split = try_get_attr(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
351 'galaxy_ml.model_validations', 'train_test_split')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
352 split_options = params['outer_split']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
353
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
354 # splits
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
355 if split_options['shuffle'] == 'stratified':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
356 split_options['labels'] = y
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
357 X, X_test, y, y_test = train_test_split(X, y, **split_options)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
358 elif split_options['shuffle'] == 'group':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
359 if groups is None:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
360 raise ValueError("No group based CV option was choosen for "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
361 "group shuffle!")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
362 split_options['labels'] = groups
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
363 if y is None:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
364 X, X_test, groups, _ =\
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
365 train_test_split(X, groups, **split_options)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
366 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
367 X, X_test, y, y_test, groups, _ =\
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
368 train_test_split(X, y, groups, **split_options)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
369 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
370 if split_options['shuffle'] == 'None':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
371 split_options['shuffle'] = None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
372 X, X_test, y, y_test =\
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
373 train_test_split(X, y, **split_options)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
374
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
375 if error_score == 'raise':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
376 searcher.fit(X, y, groups=groups)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
377 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
378 warnings.simplefilter('always', FitFailedWarning)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
379 with warnings.catch_warnings(record=True) as w:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
380 try:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
381 searcher.fit(X, y, groups=groups)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
382 except ValueError:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
383 pass
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
384 for warning in w:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
385 print(repr(warning.message))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
386
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
387 scorer_ = searcher.scorer_
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
388 if isinstance(scorer_, collections.Mapping):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
389 is_multimetric = True
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
390 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
391 is_multimetric = False
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
392
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
393 best_estimator_ = getattr(searcher, 'best_estimator_')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
394
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
395 # TODO Solve deep learning models in pipeline
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
396 if best_estimator_.__class__.__name__ == 'KerasGBatchClassifier':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
397 test_score = best_estimator_.evaluate(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
398 X_test, scorer=scorer_, is_multimetric=is_multimetric)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
399 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
400 test_score = _score(best_estimator_, X_test,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
401 y_test, scorer_,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
402 is_multimetric=is_multimetric)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
403
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
404 if not is_multimetric:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
405 test_score = {primary_scoring: test_score}
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
406 for key, value in test_score.items():
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
407 test_score[key] = [value]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
408 result_df = pd.DataFrame(test_score)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
409 result_df.to_csv(path_or_buf=outfile, sep='\t', header=True,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
410 index=False)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
411
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
412 return searcher
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
413
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
414
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
415 def main(inputs, infile_estimator, infile1, infile2,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
416 outfile_result, outfile_object=None,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
417 outfile_weights=None, groups=None,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
418 ref_seq=None, intervals=None, targets=None,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
419 fasta_path=None):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
420 """
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
421 Parameter
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
422 ---------
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
423 inputs : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
424 File path to galaxy tool parameter
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
425
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
426 infile_estimator : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
427 File path to estimator
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
428
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
429 infile1 : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
430 File path to dataset containing features
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
431
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
432 infile2 : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
433 File path to dataset containing target values
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
434
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
435 outfile_result : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
436 File path to save the results, either cv_results or test result
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
437
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
438 outfile_object : str, optional
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
439 File path to save searchCV object
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
440
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
441 outfile_weights : str, optional
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
442 File path to save model weights
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
443
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
444 groups : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
445 File path to dataset containing groups labels
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
446
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
447 ref_seq : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
448 File path to dataset containing genome sequence file
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
449
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
450 intervals : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
451 File path to dataset containing interval file
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
452
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
453 targets : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
454 File path to dataset compressed target bed file
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
455
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
456 fasta_path : str
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
457 File path to dataset containing fasta file
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
458 """
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
459 warnings.simplefilter('ignore')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
460
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
461 # store read dataframe object
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
462 loaded_df = {}
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
463
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
464 with open(inputs, 'r') as param_handler:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
465 params = json.load(param_handler)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
466
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
467 # Override the refit parameter
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
468 params['search_schemes']['options']['refit'] = True \
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
469 if params['save'] != 'nope' else False
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
470
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
471 with open(infile_estimator, 'rb') as estimator_handler:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
472 estimator = load_model(estimator_handler)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
473
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
474 optimizer = params['search_schemes']['selected_search_scheme']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
475 optimizer = getattr(model_selection, optimizer)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
476
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
477 # handle gridsearchcv options
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
478 options = params['search_schemes']['options']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
479
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
480 if groups:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
481 header = 'infer' if (options['cv_selector']['groups_selector']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
482 ['header_g']) else None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
483 column_option = (options['cv_selector']['groups_selector']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
484 ['column_selector_options_g']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
485 ['selected_column_selector_option_g'])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
486 if column_option in ['by_index_number', 'all_but_by_index_number',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
487 'by_header_name', 'all_but_by_header_name']:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
488 c = (options['cv_selector']['groups_selector']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
489 ['column_selector_options_g']['col_g'])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
490 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
491 c = None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
492
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
493 df_key = groups + repr(header)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
494
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
495 groups = pd.read_csv(groups, sep='\t', header=header,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
496 parse_dates=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
497 loaded_df[df_key] = groups
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
498
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
499 groups = read_columns(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
500 groups,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
501 c=c,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
502 c_option=column_option,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
503 sep='\t',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
504 header=header,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
505 parse_dates=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
506 groups = groups.ravel()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
507 options['cv_selector']['groups_selector'] = groups
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
508
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
509 splitter, groups = get_cv(options.pop('cv_selector'))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
510 options['cv'] = splitter
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
511 primary_scoring = options['scoring']['primary_scoring']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
512 options['scoring'] = get_scoring(options['scoring'])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
513 if options['error_score']:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
514 options['error_score'] = 'raise'
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
515 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
516 options['error_score'] = np.NaN
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
517 if options['refit'] and isinstance(options['scoring'], dict):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
518 options['refit'] = primary_scoring
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
519 if 'pre_dispatch' in options and options['pre_dispatch'] == '':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
520 options['pre_dispatch'] = None
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
521
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
522 params_builder = params['search_schemes']['search_params_builder']
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
523 param_grid = _eval_search_params(params_builder)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
524
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
525 estimator = clean_params(estimator)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
526
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
527 # save the SearchCV object without fit
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
528 if params['save'] == 'save_no_fit':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
529 searcher = optimizer(estimator, param_grid, **options)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
530 print(searcher)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
531 with open(outfile_object, 'wb') as output_handler:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
532 pickle.dump(searcher, output_handler,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
533 pickle.HIGHEST_PROTOCOL)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
534 return 0
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
535
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
536 # read inputs and loads new attributes, like paths
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
537 estimator, X, y = _handle_X_y(estimator, params, infile1, infile2,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
538 loaded_df=loaded_df, ref_seq=ref_seq,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
539 intervals=intervals, targets=targets,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
540 fasta_path=fasta_path)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
541
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
542 # cache iraps_core fits could increase search speed significantly
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
543 memory = joblib.Memory(location=CACHE_DIR, verbose=0)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
544 main_est = get_main_estimator(estimator)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
545 if main_est.__class__.__name__ == 'IRAPSClassifier':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
546 main_est.set_params(memory=memory)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
547
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
548 searcher = optimizer(estimator, param_grid, **options)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
549
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
550 split_mode = params['outer_split'].pop('split_mode')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
551
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
552 if split_mode == 'nested_cv':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
553 # make sure refit is choosen
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
554 # this could be True for sklearn models, but not the case for
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
555 # deep learning models
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
556 if not options['refit'] and \
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
557 not all(hasattr(estimator, attr)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
558 for attr in ('config', 'model_type')):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
559 warnings.warn("Refit is change to `True` for nested validation!")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
560 setattr(searcher, 'refit', True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
561
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
562 outer_cv, _ = get_cv(params['outer_split']['cv_selector'])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
563 # nested CV, outer cv using cross_validate
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
564 if options['error_score'] == 'raise':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
565 rval = cross_validate(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
566 searcher, X, y, scoring=options['scoring'],
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
567 cv=outer_cv, n_jobs=N_JOBS,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
568 verbose=options['verbose'],
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
569 return_estimator=(params['save'] == 'save_estimator'),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
570 error_score=options['error_score'],
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
571 return_train_score=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
572 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
573 warnings.simplefilter('always', FitFailedWarning)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
574 with warnings.catch_warnings(record=True) as w:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
575 try:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
576 rval = cross_validate(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
577 searcher, X, y,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
578 scoring=options['scoring'],
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
579 cv=outer_cv, n_jobs=N_JOBS,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
580 verbose=options['verbose'],
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
581 return_estimator=(params['save'] == 'save_estimator'),
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
582 error_score=options['error_score'],
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
583 return_train_score=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
584 except ValueError:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
585 pass
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
586 for warning in w:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
587 print(repr(warning.message))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
588
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
589 fitted_searchers = rval.pop('estimator', [])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
590 if fitted_searchers:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
591 import os
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
592 pwd = os.getcwd()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
593 save_dir = os.path.join(pwd, 'cv_results_in_folds')
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
594 try:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
595 os.mkdir(save_dir)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
596 for idx, obj in enumerate(fitted_searchers):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
597 target_name = 'cv_results_' + '_' + 'split%d' % idx
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
598 target_path = os.path.join(pwd, save_dir, target_name)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
599 cv_results_ = getattr(obj, 'cv_results_', None)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
600 if not cv_results_:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
601 print("%s is not available" % target_name)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
602 continue
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
603 cv_results_ = pd.DataFrame(cv_results_)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
604 cv_results_ = cv_results_[sorted(cv_results_.columns)]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
605 cv_results_.to_csv(target_path, sep='\t', header=True,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
606 index=False)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
607 except Exception as e:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
608 print(e)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
609 finally:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
610 del os
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
611
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
612 keys = list(rval.keys())
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
613 for k in keys:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
614 if k.startswith('test'):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
615 rval['mean_' + k] = np.mean(rval[k])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
616 rval['std_' + k] = np.std(rval[k])
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
617 if k.endswith('time'):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
618 rval.pop(k)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
619 rval = pd.DataFrame(rval)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
620 rval = rval[sorted(rval.columns)]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
621 rval.to_csv(path_or_buf=outfile_result, sep='\t', header=True,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
622 index=False)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
623
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
624 return 0
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
625
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
626 # deprecate train test split mode
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
627 """searcher = _do_train_test_split_val(
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
628 searcher, X, y, params,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
629 primary_scoring=primary_scoring,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
630 error_score=options['error_score'],
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
631 groups=groups,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
632 outfile=outfile_result)"""
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
633
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
634 # no outer split
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
635 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
636 searcher.set_params(n_jobs=N_JOBS)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
637 if options['error_score'] == 'raise':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
638 searcher.fit(X, y, groups=groups)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
639 else:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
640 warnings.simplefilter('always', FitFailedWarning)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
641 with warnings.catch_warnings(record=True) as w:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
642 try:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
643 searcher.fit(X, y, groups=groups)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
644 except ValueError:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
645 pass
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
646 for warning in w:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
647 print(repr(warning.message))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
648
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
649 cv_results = pd.DataFrame(searcher.cv_results_)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
650 cv_results = cv_results[sorted(cv_results.columns)]
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
651 cv_results.to_csv(path_or_buf=outfile_result, sep='\t',
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
652 header=True, index=False)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
653
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
654 memory.clear(warn=False)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
655
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
656 # output best estimator, and weights if applicable
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
657 if outfile_object:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
658 best_estimator_ = getattr(searcher, 'best_estimator_', None)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
659 if not best_estimator_:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
660 warnings.warn("GridSearchCV object has no attribute "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
661 "'best_estimator_', because either it's "
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
662 "nested gridsearch or `refit` is False!")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
663 return
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
664
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
665 # clean prams
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
666 best_estimator_ = clean_params(best_estimator_)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
667
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
668 main_est = get_main_estimator(best_estimator_)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
669
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
670 if hasattr(main_est, 'model_') \
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
671 and hasattr(main_est, 'save_weights'):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
672 if outfile_weights:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
673 main_est.save_weights(outfile_weights)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
674 del main_est.model_
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
675 del main_est.fit_params
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
676 del main_est.model_class_
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
677 del main_est.validation_data
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
678 if getattr(main_est, 'data_generator_', None):
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
679 del main_est.data_generator_
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
680
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
681 with open(outfile_object, 'wb') as output_handler:
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
682 print("Best estimator is saved: %s " % repr(best_estimator_))
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
683 pickle.dump(best_estimator_, output_handler,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
684 pickle.HIGHEST_PROTOCOL)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
685
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
686
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
687 if __name__ == '__main__':
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
688 aparser = argparse.ArgumentParser()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
689 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
690 aparser.add_argument("-e", "--estimator", dest="infile_estimator")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
691 aparser.add_argument("-X", "--infile1", dest="infile1")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
692 aparser.add_argument("-y", "--infile2", dest="infile2")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
693 aparser.add_argument("-O", "--outfile_result", dest="outfile_result")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
694 aparser.add_argument("-o", "--outfile_object", dest="outfile_object")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
695 aparser.add_argument("-w", "--outfile_weights", dest="outfile_weights")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
696 aparser.add_argument("-g", "--groups", dest="groups")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
697 aparser.add_argument("-r", "--ref_seq", dest="ref_seq")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
698 aparser.add_argument("-b", "--intervals", dest="intervals")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
699 aparser.add_argument("-t", "--targets", dest="targets")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
700 aparser.add_argument("-f", "--fasta_path", dest="fasta_path")
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
701 args = aparser.parse_args()
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
702
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
703 main(args.inputs, args.infile_estimator, args.infile1, args.infile2,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
704 args.outfile_result, outfile_object=args.outfile_object,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
705 outfile_weights=args.outfile_weights, groups=args.groups,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
706 ref_seq=args.ref_seq, intervals=args.intervals,
03f61bb3ca43 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
707 targets=args.targets, fasta_path=args.fasta_path)