sklearn_stacking_ensemble_models: search_model

annotate search_model_validation.py @ 0:47467890f541 draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c

author	bgruening
date	Wed, 15 May 2019 07:03:47 -0400
parents
children	e18d9b17c322

rev	line source
0 47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	1 import argparse
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	2 import collections
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	3 import imblearn
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	4 import json
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	5 import numpy as np
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	6 import pandas
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	7 import pickle
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	8 import skrebate
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	9 import sklearn
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	10 import sys
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	11 import xgboost
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	12 import warnings
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	13 import iraps_classifier
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	14 import model_validations
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	15 import preprocessors
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	16 import feature_selectors
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	17 from imblearn import under_sampling, over_sampling, combine
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	18 from scipy.io import mmread
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	19 from mlxtend import classifier, regressor
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	20 from sklearn import (cluster, compose, decomposition, ensemble,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	21 feature_extraction, feature_selection,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	22 gaussian_process, kernel_approximation, metrics,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	23 model_selection, naive_bayes, neighbors,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	24 pipeline, preprocessing, svm, linear_model,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	25 tree, discriminant_analysis)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	26 from sklearn.exceptions import FitFailedWarning
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	27 from sklearn.externals import joblib
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	28 from sklearn.model_selection._validation import _score
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	29
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	30 from utils import (SafeEval, get_cv, get_scoring, get_X_y,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	31 load_model, read_columns)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	32 from model_validations import train_test_split
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	33
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	34
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	35 N_JOBS = int(__import__('os').environ.get('GALAXY_SLOTS', 1))
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	36 CACHE_DIR = './cached'
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	37 NON_SEARCHABLE = ('n_jobs', 'pre_dispatch', 'memory', 'steps',
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	38 'nthread', 'verbose')
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	39
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	40
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	41 def _eval_search_params(params_builder):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	42 search_params = {}
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	43
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	44 for p in params_builder['param_set']:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	45 search_list = p['sp_list'].strip()
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	46 if search_list == '':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	47 continue
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	48
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	49 param_name = p['sp_name']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	50 if param_name.lower().endswith(NON_SEARCHABLE):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	51 print("Warning: `%s` is not eligible for search and was "
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	52 "omitted!" % param_name)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	53 continue
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	54
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	55 if not search_list.startswith(':'):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	56 safe_eval = SafeEval(load_scipy=True, load_numpy=True)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	57 ev = safe_eval(search_list)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	58 search_params[param_name] = ev
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	59 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	60 # Have `:` before search list, asks for estimator evaluatio
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	61 safe_eval_es = SafeEval(load_estimators=True)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	62 search_list = search_list[1:].strip()
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	63 # TODO maybe add regular express check
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	64 ev = safe_eval_es(search_list)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	65 preprocessors = (
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	66 preprocessing.StandardScaler(), preprocessing.Binarizer(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	67 preprocessing.Imputer(), preprocessing.MaxAbsScaler(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	68 preprocessing.Normalizer(), preprocessing.MinMaxScaler(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	69 preprocessing.PolynomialFeatures(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	70 preprocessing.RobustScaler(), feature_selection.SelectKBest(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	71 feature_selection.GenericUnivariateSelect(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	72 feature_selection.SelectPercentile(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	73 feature_selection.SelectFpr(), feature_selection.SelectFdr(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	74 feature_selection.SelectFwe(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	75 feature_selection.VarianceThreshold(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	76 decomposition.FactorAnalysis(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	77 decomposition.FastICA(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	78 decomposition.IncrementalPCA(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	79 decomposition.KernelPCA(random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	80 decomposition.LatentDirichletAllocation(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	81 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	82 decomposition.MiniBatchDictionaryLearning(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	83 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	84 decomposition.MiniBatchSparsePCA(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	85 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	86 decomposition.NMF(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	87 decomposition.PCA(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	88 decomposition.SparsePCA(random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	89 decomposition.TruncatedSVD(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	90 kernel_approximation.Nystroem(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	91 kernel_approximation.RBFSampler(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	92 kernel_approximation.AdditiveChi2Sampler(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	93 kernel_approximation.SkewedChi2Sampler(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	94 cluster.FeatureAgglomeration(),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	95 skrebate.ReliefF(n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	96 skrebate.SURF(n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	97 skrebate.SURFstar(n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	98 skrebate.MultiSURF(n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	99 skrebate.MultiSURFstar(n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	100 imblearn.under_sampling.ClusterCentroids(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	101 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	102 imblearn.under_sampling.CondensedNearestNeighbour(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	103 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	104 imblearn.under_sampling.EditedNearestNeighbours(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	105 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	106 imblearn.under_sampling.RepeatedEditedNearestNeighbours(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	107 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	108 imblearn.under_sampling.AllKNN(random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	109 imblearn.under_sampling.InstanceHardnessThreshold(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	110 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	111 imblearn.under_sampling.NearMiss(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	112 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	113 imblearn.under_sampling.NeighbourhoodCleaningRule(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	114 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	115 imblearn.under_sampling.OneSidedSelection(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	116 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	117 imblearn.under_sampling.RandomUnderSampler(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	118 random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	119 imblearn.under_sampling.TomekLinks(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	120 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	121 imblearn.over_sampling.ADASYN(random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	122 imblearn.over_sampling.RandomOverSampler(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	123 imblearn.over_sampling.SMOTE(random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	124 imblearn.over_sampling.SVMSMOTE(random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	125 imblearn.over_sampling.BorderlineSMOTE(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	126 random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	127 imblearn.over_sampling.SMOTENC(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	128 categorical_features=[], random_state=0, n_jobs=N_JOBS),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	129 imblearn.combine.SMOTEENN(random_state=0),
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	130 imblearn.combine.SMOTETomek(random_state=0))
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	131 newlist = []
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	132 for obj in ev:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	133 if obj is None:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	134 newlist.append(None)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	135 elif obj == 'all_0':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	136 newlist.extend(preprocessors[0:36])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	137 elif obj == 'sk_prep_all': # no KernalCenter()
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	138 newlist.extend(preprocessors[0:8])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	139 elif obj == 'fs_all':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	140 newlist.extend(preprocessors[8:15])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	141 elif obj == 'decomp_all':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	142 newlist.extend(preprocessors[15:26])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	143 elif obj == 'k_appr_all':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	144 newlist.extend(preprocessors[26:30])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	145 elif obj == 'reb_all':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	146 newlist.extend(preprocessors[31:36])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	147 elif obj == 'imb_all':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	148 newlist.extend(preprocessors[36:55])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	149 elif type(obj) is int and -1 < obj < len(preprocessors):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	150 newlist.append(preprocessors[obj])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	151 elif hasattr(obj, 'get_params'): # user uploaded object
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	152 if 'n_jobs' in obj.get_params():
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	153 newlist.append(obj.set_params(n_jobs=N_JOBS))
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	154 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	155 newlist.append(obj)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	156 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	157 sys.exit("Unsupported estimator type: %r" % (obj))
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	158
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	159 search_params[param_name] = newlist
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	160
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	161 return search_params
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	162
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	163
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	164 def main(inputs, infile_estimator, infile1, infile2,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	165 outfile_result, outfile_object=None, groups=None):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	166 """
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	167 Parameter
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	168 ---------
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	169 inputs : str
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	170 File path to galaxy tool parameter
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	171
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	172 infile_estimator : str
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	173 File path to estimator
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	174
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	175 infile1 : str
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	176 File path to dataset containing features
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	177
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	178 infile2 : str
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	179 File path to dataset containing target values
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	180
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	181 outfile_result : str
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	182 File path to save the results, either cv_results or test result
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	183
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	184 outfile_object : str, optional
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	185 File path to save searchCV object
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	186
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	187 groups : str
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	188 File path to dataset containing groups labels
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	189 """
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	190
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	191 warnings.simplefilter('ignore')
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	192
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	193 with open(inputs, 'r') as param_handler:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	194 params = json.load(param_handler)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	195 if groups:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	196 (params['search_schemes']['options']['cv_selector']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	197 ['groups_selector']['infile_g']) = groups
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	198
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	199 params_builder = params['search_schemes']['search_params_builder']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	200
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	201 input_type = params['input_options']['selected_input']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	202 if input_type == 'tabular':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	203 header = 'infer' if params['input_options']['header1'] else None
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	204 column_option = (params['input_options']['column_selector_options_1']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	205 ['selected_column_selector_option'])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	206 if column_option in ['by_index_number', 'all_but_by_index_number',
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	207 'by_header_name', 'all_but_by_header_name']:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	208 c = params['input_options']['column_selector_options_1']['col1']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	209 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	210 c = None
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	211 X = read_columns(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	212 infile1,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	213 c=c,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	214 c_option=column_option,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	215 sep='\t',
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	216 header=header,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	217 parse_dates=True).astype(float)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	218 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	219 X = mmread(open(infile1, 'r'))
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	220
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	221 header = 'infer' if params['input_options']['header2'] else None
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	222 column_option = (params['input_options']['column_selector_options_2']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	223 ['selected_column_selector_option2'])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	224 if column_option in ['by_index_number', 'all_but_by_index_number',
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	225 'by_header_name', 'all_but_by_header_name']:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	226 c = params['input_options']['column_selector_options_2']['col2']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	227 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	228 c = None
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	229 y = read_columns(
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	230 infile2,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	231 c=c,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	232 c_option=column_option,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	233 sep='\t',
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	234 header=header,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	235 parse_dates=True)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	236 y = y.ravel()
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	237
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	238 optimizer = params['search_schemes']['selected_search_scheme']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	239 optimizer = getattr(model_selection, optimizer)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	240
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	241 options = params['search_schemes']['options']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	242
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	243 splitter, groups = get_cv(options.pop('cv_selector'))
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	244 options['cv'] = splitter
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	245 options['n_jobs'] = N_JOBS
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	246 primary_scoring = options['scoring']['primary_scoring']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	247 options['scoring'] = get_scoring(options['scoring'])
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	248 if options['error_score']:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	249 options['error_score'] = 'raise'
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	250 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	251 options['error_score'] = np.NaN
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	252 if options['refit'] and isinstance(options['scoring'], dict):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	253 options['refit'] = primary_scoring
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	254 if 'pre_dispatch' in options and options['pre_dispatch'] == '':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	255 options['pre_dispatch'] = None
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	256
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	257 with open(infile_estimator, 'rb') as estimator_handler:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	258 estimator = load_model(estimator_handler)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	259
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	260 memory = joblib.Memory(location=CACHE_DIR, verbose=0)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	261 # cache iraps_core fits could increase search speed significantly
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	262 if estimator.__class__.__name__ == 'IRAPSClassifier':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	263 estimator.set_params(memory=memory)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	264 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	265 for p, v in estimator.get_params().items():
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	266 if p.endswith('memory'):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	267 if len(p) > 8 and p[:-8].endswith('irapsclassifier'):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	268 # cache iraps_core fits could increase search
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	269 # speed significantly
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	270 new_params = {p: memory}
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	271 estimator.set_params(**new_params)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	272 elif v:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	273 new_params = {p, None}
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	274 estimator.set_params(**new_params)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	275 elif p.endswith('n_jobs'):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	276 new_params = {p: 1}
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	277 estimator.set_params(**new_params)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	278
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	279 param_grid = _eval_search_params(params_builder)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	280 searcher = optimizer(estimator, param_grid, **options)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	281
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	282 # do train_test_split
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	283 do_train_test_split = params['train_test_split'].pop('do_split')
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	284 if do_train_test_split == 'yes':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	285 # make sure refit is choosen
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	286 if not options['refit']:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	287 raise ValueError("Refit must be `True` for shuffle splitting!")
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	288 split_options = params['train_test_split']
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	289
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	290 # splits
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	291 if split_options['shuffle'] == 'stratified':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	292 split_options['labels'] = y
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	293 X, X_test, y, y_test = train_test_split(X, y, **split_options)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	294 elif split_options['shuffle'] == 'group':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	295 if not groups:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	296 raise ValueError("No group based CV option was "
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	297 "choosen for group shuffle!")
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	298 split_options['labels'] = groups
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	299 X, X_test, y, y_test, groups, _ =\
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	300 train_test_split(X, y, **split_options)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	301 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	302 if split_options['shuffle'] == 'None':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	303 split_options['shuffle'] = None
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	304 X, X_test, y, y_test =\
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	305 train_test_split(X, y, **split_options)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	306 # end train_test_split
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	307
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	308 if options['error_score'] == 'raise':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	309 searcher.fit(X, y, groups=groups)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	310 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	311 warnings.simplefilter('always', FitFailedWarning)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	312 with warnings.catch_warnings(record=True) as w:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	313 try:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	314 searcher.fit(X, y, groups=groups)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	315 except ValueError:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	316 pass
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	317 for warning in w:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	318 print(repr(warning.message))
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	319
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	320 if do_train_test_split == 'no':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	321 # save results
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	322 cv_results = pandas.DataFrame(searcher.cv_results_)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	323 cv_results = cv_results[sorted(cv_results.columns)]
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	324 cv_results.to_csv(path_or_buf=outfile_result, sep='\t',
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	325 header=True, index=False)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	326
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	327 # output test result using best_estimator_
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	328 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	329 best_estimator_ = searcher.best_estimator_
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	330 if isinstance(options['scoring'], collections.Mapping):
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	331 is_multimetric = True
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	332 else:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	333 is_multimetric = False
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	334
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	335 test_score = _score(best_estimator_, X_test,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	336 y_test, options['scoring'],
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	337 is_multimetric=is_multimetric)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	338 if not is_multimetric:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	339 test_score = {primary_scoring: test_score}
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	340 for key, value in test_score.items():
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	341 test_score[key] = [value]
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	342 result_df = pandas.DataFrame(test_score)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	343 result_df.to_csv(path_or_buf=outfile_result, sep='\t',
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	344 header=True, index=False)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	345
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	346 memory.clear(warn=False)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	347
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	348 if outfile_object:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	349 with open(outfile_object, 'wb') as output_handler:
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	350 pickle.dump(searcher, output_handler, pickle.HIGHEST_PROTOCOL)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	351
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	352
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	353 if __name__ == '__main__':
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	354 aparser = argparse.ArgumentParser()
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	355 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	356 aparser.add_argument("-e", "--estimator", dest="infile_estimator")
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	357 aparser.add_argument("-X", "--infile1", dest="infile1")
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	358 aparser.add_argument("-y", "--infile2", dest="infile2")
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	359 aparser.add_argument("-r", "--outfile_result", dest="outfile_result")
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	360 aparser.add_argument("-o", "--outfile_object", dest="outfile_object")
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	361 aparser.add_argument("-g", "--groups", dest="groups")
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	362 args = aparser.parse_args()
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	363
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	364 main(args.inputs, args.infile_estimator, args.infile1, args.infile2,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	365 args.outfile_result, outfile_object=args.outfile_object,
47467890f541 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ab963ec9498bd05d2fb2f24f75adb2fccae7958c bgruening parents: diff changeset	366 groups=args.groups)

Mercurial > repos > bgruening > sklearn_stacking_ensemble_models

annotate search_model_validation.py @ 0:47467890f541 draft