Mercurial > repos > bgruening > sklearn_numeric_clustering

annotate utils.py @ 21:1d6d938e7ee2 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
author bgruening
date Sat, 29 Sep 2018 07:29:32 -0400
parents e6b45e6447fc
children 9d234733ccfd
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
1 import sys
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
2 import os
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
3 import pandas
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
4 import re
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
5 import pickle
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
6 import warnings
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
7 import numpy as np
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
8 import xgboost
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
9 import scipy
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
10 import sklearn
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
11 from asteval import Interpreter, make_symbol_table
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
12 from sklearn import (cluster, decomposition, ensemble, feature_extraction, feature_selection,
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
13 gaussian_process, kernel_approximation, metrics,
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
14 model_selection, naive_bayes, neighbors, pipeline, preprocessing,
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
15 svm, linear_model, tree, discriminant_analysis)
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
16
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
17 N_JOBS = int(os.environ.get('GALAXY_SLOTS', 1))
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
18
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
19
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
20 class SafePickler(pickle.Unpickler):
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
21 """
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
22 Used to safely deserialize scikit-learn model objects serialized by cPickle.dump
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
23 Usage:
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
24 eg.: SafePickler.load(pickled_file_object)
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
25 """
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
26 def find_class(self, module, name):
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
27
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
28 bad_names = ('and', 'as', 'assert', 'break', 'class', 'continue',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
29 'def', 'del', 'elif', 'else', 'except', 'exec',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
30 'finally', 'for', 'from', 'global', 'if', 'import',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
31 'in', 'is', 'lambda', 'not', 'or', 'pass', 'print',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
32 'raise', 'return', 'try', 'system', 'while', 'with',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
33 'True', 'False', 'None', 'eval', 'execfile', '__import__',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
34 '__package__', '__subclasses__', '__bases__', '__globals__',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
35 '__code__', '__closure__', '__func__', '__self__', '__module__',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
36 '__dict__', '__class__', '__call__', '__get__',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
37 '__getattribute__', '__subclasshook__', '__new__',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
38 '__init__', 'func_globals', 'func_code', 'func_closure',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
39 'im_class', 'im_func', 'im_self', 'gi_code', 'gi_frame',
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
40 '__asteval__', 'f_locals', '__mro__')
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
41 good_names = ['copy_reg._reconstructor', '__builtin__.object']
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
42
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
43 if re.match(r'^[a-zA-Z_][a-zA-Z0-9_]*$', name):
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
44 fullname = module + '.' + name
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
45 if (fullname in good_names)\
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
46 or ( ( module.startswith('sklearn.')
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
47 or module.startswith('xgboost.')
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
48 or module.startswith('skrebate.')
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
49 or module.startswith('numpy.')
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
50 or module == 'numpy'
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
51 )
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
52 and (name not in bad_names)
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
53 ):
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
54 # TODO: replace with a whitelist checker
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
55 if fullname not in sk_whitelist['SK_NAMES'] + sk_whitelist['SKR_NAMES'] + sk_whitelist['XGB_NAMES'] + sk_whitelist['NUMPY_NAMES'] + good_names:
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
56 print("Warning: global %s is not in pickler whitelist yet and will loss support soon. Contact tool author or leave a message at github.com" % fullname)
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
57 mod = sys.modules[module]
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
58 return getattr(mod, name)
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
59
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
60 raise pickle.UnpicklingError("global '%s' is forbidden" % fullname)
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
61
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
62
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
63 def load_model(file):
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
64 return SafePickler(file).load()
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
65
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
66
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
67 def read_columns(f, c=None, c_option='by_index_number', return_df=False, **args):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
68 data = pandas.read_csv(f, **args)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
69 if c_option == 'by_index_number':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
70 cols = list(map(lambda x: x - 1, c))
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
71 data = data.iloc[:, cols]
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
72 if c_option == 'all_but_by_index_number':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
73 cols = list(map(lambda x: x - 1, c))
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
74 data.drop(data.columns[cols], axis=1, inplace=True)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
75 if c_option == 'by_header_name':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
76 cols = [e.strip() for e in c.split(',')]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
77 data = data[cols]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
78 if c_option == 'all_but_by_header_name':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
79 cols = [e.strip() for e in c.split(',')]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
80 data.drop(cols, axis=1, inplace=True)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
81 y = data.values
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
82 if return_df:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
83 return y, data
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
84 else:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
85 return y
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
86 return y
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
87
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
88
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
89 ## generate an instance for one of sklearn.feature_selection classes
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
90 def feature_selector(inputs):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
91 selector = inputs["selected_algorithm"]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
92 selector = getattr(sklearn.feature_selection, selector)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
93 options = inputs["options"]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
94
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
95 if inputs['selected_algorithm'] == 'SelectFromModel':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
96 if not options['threshold'] or options['threshold'] == 'None':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
97 options['threshold'] = None
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
98 if inputs['model_inputter']['input_mode'] == 'prefitted':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
99 model_file = inputs['model_inputter']['fitted_estimator']
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
100 with open(model_file, 'rb') as model_handler:
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
101 fitted_estimator = load_model(model_handler)
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
102 new_selector = selector(fitted_estimator, prefit=True, **options)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
103 else:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
104 estimator_json = inputs['model_inputter']["estimator_selector"]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
105 estimator = get_estimator(estimator_json)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
106 new_selector = selector(estimator, **options)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
107
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
108 elif inputs['selected_algorithm'] == 'RFE':
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
109 estimator = get_estimator(inputs["estimator_selector"])
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
110 new_selector = selector(estimator, **options)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
111
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
112 elif inputs['selected_algorithm'] == 'RFECV':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
113 options['scoring'] = get_scoring(options['scoring'])
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
114 options['n_jobs'] = N_JOBS
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
115 options['cv'] = get_cv(options['cv'].strip())
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
116 estimator = get_estimator(inputs["estimator_selector"])
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
117 new_selector = selector(estimator, **options)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
118
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
119 elif inputs['selected_algorithm'] == "VarianceThreshold":
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
120 new_selector = selector(**options)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
121
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
122 else:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
123 score_func = inputs["score_func"]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
124 score_func = getattr(sklearn.feature_selection, score_func)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
125 new_selector = selector(score_func, **options)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
126
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
127 return new_selector
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
128
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
129
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
130 def get_X_y(params, file1, file2):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
131 input_type = params["selected_tasks"]["selected_algorithms"]["input_options"]["selected_input"]
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
132 if input_type == "tabular":
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
133 header = 'infer' if params["selected_tasks"]["selected_algorithms"]["input_options"]["header1"] else None
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
134 column_option = params["selected_tasks"]["selected_algorithms"]["input_options"]["column_selector_options_1"]["selected_column_selector_option"]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
135 if column_option in ["by_index_number", "all_but_by_index_number", "by_header_name", "all_but_by_header_name"]:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
136 c = params["selected_tasks"]["selected_algorithms"]["input_options"]["column_selector_options_1"]["col1"]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
137 else:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
138 c = None
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
139 X = read_columns(
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
140 file1,
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
141 c=c,
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
142 c_option=column_option,
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
143 sep='\t',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
144 header=header,
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
145 parse_dates=True
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
146 )
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
147 else:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
148 X = mmread(file1)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
149
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
150 header = 'infer' if params["selected_tasks"]["selected_algorithms"]["input_options"]["header2"] else None
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
151 column_option = params["selected_tasks"]["selected_algorithms"]["input_options"]["column_selector_options_2"]["selected_column_selector_option2"]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
152 if column_option in ["by_index_number", "all_but_by_index_number", "by_header_name", "all_but_by_header_name"]:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
153 c = params["selected_tasks"]["selected_algorithms"]["input_options"]["column_selector_options_2"]["col2"]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
154 else:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
155 c = None
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
156 y = read_columns(
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
157 file2,
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
158 c=c,
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
159 c_option=column_option,
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
160 sep='\t',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
161 header=header,
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
162 parse_dates=True
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
163 )
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
164 y = y.ravel()
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
165 return X, y
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
166
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
167
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
168 class SafeEval(Interpreter):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
169
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
170 def __init__(self, load_scipy=False, load_numpy=False):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
171
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
172 # File opening and other unneeded functions could be dropped
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
173 unwanted = ['open', 'type', 'dir', 'id', 'str', 'repr']
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
174
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
175 # Allowed symbol table. Add more if needed.
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
176 new_syms = {
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
177 'np_arange': getattr(np, 'arange'),
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
178 'ensemble_ExtraTreesClassifier': getattr(ensemble, 'ExtraTreesClassifier')
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
179 }
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
180
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
181 syms = make_symbol_table(use_numpy=False, **new_syms)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
182
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
183 if load_scipy:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
184 scipy_distributions = scipy.stats.distributions.__dict__
20
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
185 for k, v in scipy_distributions.items():
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
186 if isinstance(v, (scipy.stats.rv_continuous, scipy.stats.rv_discrete)):
e6b45e6447fc planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8cf3d813ec755166ee0bd517b4ecbbd4f84d4df1
bgruening
parents: 19
diff changeset
187 syms['scipy_stats_' + k] = v
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
188
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
189 if load_numpy:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
190 from_numpy_random = ['beta', 'binomial', 'bytes', 'chisquare', 'choice', 'dirichlet', 'division',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
191 'exponential', 'f', 'gamma', 'geometric', 'gumbel', 'hypergeometric',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
192 'laplace', 'logistic', 'lognormal', 'logseries', 'mtrand', 'multinomial',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
193 'multivariate_normal', 'negative_binomial', 'noncentral_chisquare', 'noncentral_f',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
194 'normal', 'pareto', 'permutation', 'poisson', 'power', 'rand', 'randint',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
195 'randn', 'random', 'random_integers', 'random_sample', 'ranf', 'rayleigh',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
196 'sample', 'seed', 'set_state', 'shuffle', 'standard_cauchy', 'standard_exponential',
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
197 'standard_gamma', 'standard_normal', 'standard_t', 'triangular', 'uniform',
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
198 'vonmises', 'wald', 'weibull', 'zipf']
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
199 for f in from_numpy_random:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
200 syms['np_random_' + f] = getattr(np.random, f)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
201
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
202 for key in unwanted:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
203 syms.pop(key, None)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
204
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
205 super(SafeEval, self).__init__(symtable=syms, use_numpy=False, minimal=False,
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
206 no_if=True, no_for=True, no_while=True, no_try=True,
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
207 no_functiondef=True, no_ifexp=True, no_listcomp=False,
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
208 no_augassign=False, no_assert=True, no_delete=True,
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
209 no_raise=True, no_print=True)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
210
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
211
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
212 def get_search_params(params_builder):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
213 search_params = {}
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
214 safe_eval = SafeEval(load_scipy=True, load_numpy=True)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
215
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
216 for p in params_builder['param_set']:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
217 search_p = p['search_param_selector']['search_p']
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
218 if search_p.strip() == '':
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
219 continue
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
220 param_type = p['search_param_selector']['selected_param_type']
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
221
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
222 lst = search_p.split(":")
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
223 assert (len(lst) == 2), "Error, make sure there is one and only one colon in search parameter input."
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
224 literal = lst[1].strip()
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
225 ev = safe_eval(literal)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
226 if param_type == "final_estimator_p":
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
227 search_params["estimator__" + lst[0].strip()] = ev
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
228 else:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
229 search_params["preprocessing_" + param_type[5:6] + "__" + lst[0].strip()] = ev
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
230
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
231 return search_params
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
232
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
233
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
234 def get_estimator(estimator_json):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
235 estimator_module = estimator_json['selected_module']
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
236 estimator_cls = estimator_json['selected_estimator']
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
237
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
238 if estimator_module == "xgboost":
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
239 cls = getattr(xgboost, estimator_cls)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
240 else:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
241 module = getattr(sklearn, estimator_module)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
242 cls = getattr(module, estimator_cls)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
243
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
244 estimator = cls()
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
245
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
246 estimator_params = estimator_json['text_params'].strip()
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
247 if estimator_params != "":
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
248 try:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
249 params = safe_eval('dict(' + estimator_params + ')')
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
250 except ValueError:
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
251 sys.exit("Unsupported parameter input: `%s`" % estimator_params)
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
252 estimator.set_params(**params)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
253 if 'n_jobs' in estimator.get_params():
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
254 estimator.set_params(n_jobs=N_JOBS)
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
255
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
256 return estimator
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
257
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
258
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
259 def get_cv(literal):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
260 safe_eval = SafeEval()
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
261 if literal == "":
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
262 return None
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
263 if literal.isdigit():
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
264 return int(literal)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
265 m = re.match(r'^(?P<method>\w+)\((?P<args>.*)\)$', literal)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
266 if m:
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
267 my_class = getattr(model_selection, m.group('method'))
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
268 args = safe_eval('dict('+ m.group('args') + ')')
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
269 return my_class(**args)
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
270 sys.exit("Unsupported CV input: %s" % literal)
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
271
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
272
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
273 def get_scoring(scoring_json):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
274 def balanced_accuracy_score(y_true, y_pred):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
275 C = metrics.confusion_matrix(y_true, y_pred)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
276 with np.errstate(divide='ignore', invalid='ignore'):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
277 per_class = np.diag(C) / C.sum(axis=1)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
278 if np.any(np.isnan(per_class)):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
279 warnings.warn('y_pred contains classes not in y_true')
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
280 per_class = per_class[~np.isnan(per_class)]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
281 score = np.mean(per_class)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
282 return score
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
283
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
284 if scoring_json['primary_scoring'] == "default":
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
285 return None
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
286
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
287 my_scorers = metrics.SCORERS
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
288 if 'balanced_accuracy' not in my_scorers:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
289 my_scorers['balanced_accuracy'] = metrics.make_scorer(balanced_accuracy_score)
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
290
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
291 if scoring_json['secondary_scoring'] != 'None'\
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
292 and scoring_json['secondary_scoring'] != scoring_json['primary_scoring']:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
293 scoring = {}
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
294 scoring['primary'] = my_scorers[scoring_json['primary_scoring']]
19
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
295 for scorer in scoring_json['secondary_scoring'].split(','):
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
296 if scorer != scoring_json['primary_scoring']:
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
297 scoring[scorer] = my_scorers[scorer]
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
298 return scoring
6ba8d7af7e7a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d00173591e4a783a4c1cb2664e4bb192ab5414f7
bgruening
parents:
diff changeset
299
21
1d6d938e7ee2 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2a058459e6daf0486871f93845f00fdb4a4eaca1
bgruening
parents: 20
diff changeset
300 return my_scorers[scoring_json['primary_scoring']]