sklearn_searchcv: iraps_classifier.py annotate

annotate iraps_classifier.py @ 8:1c4a241bef5c draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7

author	bgruening
date	Tue, 14 May 2019 18:05:43 -0400
parents
children

rev	line source
8 1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	1 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	2 class IRAPSCore
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	3 class IRAPSClassifier
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	4 class BinarizeTargetClassifier
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	5 class BinarizeTargetRegressor
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	6 class _BinarizeTargetScorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	7 class _BinarizeTargetProbaScorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	8
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	9 binarize_auc_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	10 binarize_average_precision_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	11
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	12 binarize_accuracy_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	13 binarize_balanced_accuracy_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	14 binarize_precision_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	15 binarize_recall_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	16 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	17
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	18
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	19 import numpy as np
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	20 import random
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	21 import warnings
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	22
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	23 from abc import ABCMeta
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	24 from scipy.stats import ttest_ind
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	25 from sklearn import metrics
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	26 from sklearn.base import BaseEstimator, clone, RegressorMixin
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	27 from sklearn.externals import six
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	28 from sklearn.feature_selection.univariate_selection import _BaseFilter
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	29 from sklearn.metrics.scorer import _BaseScorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	30 from sklearn.pipeline import Pipeline
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	31 from sklearn.utils import as_float_array, check_X_y
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	32 from sklearn.utils._joblib import Parallel, delayed
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	33 from sklearn.utils.validation import (check_array, check_is_fitted,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	34 check_memory, column_or_1d)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	35
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	36
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	37 VERSION = '0.1.1'
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	38
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	39
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	40 class IRAPSCore(six.with_metaclass(ABCMeta, BaseEstimator)):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	41 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	42 Base class of IRAPSClassifier
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	43 From sklearn BaseEstimator:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	44 get_params()
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	45 set_params()
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	46
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	47 Parameters
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	48 ----------
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	49 n_iter : int
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	50 sample count
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	51
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	52 positive_thres : float
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	53 z_score shreshold to discretize positive target values
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	54
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	55 negative_thres : float
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	56 z_score threshold to discretize negative target values
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	57
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	58 verbose : int
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	59 0 or geater, if not 0, print progress
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	60
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	61 n_jobs : int, default=1
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	62 The number of CPUs to use to do the computation.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	63
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	64 pre_dispatch : int, or string.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	65 Controls the number of jobs that get dispatched during parallel
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	66 execution. Reducing this number can be useful to avoid an
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	67 explosion of memory consumption when more jobs get dispatched
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	68 than CPUs can process. This parameter can be:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	69 - None, in which case all the jobs are immediately
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	70 created and spawned. Use this for lightweight and
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	71 fast-running jobs, to avoid delays due to on-demand
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	72 spawning of the jobs
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	73 - An int, giving the exact number of total jobs that are
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	74 spawned
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	75 - A string, giving an expression as a function of n_jobs,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	76 as in '2*n_jobs'
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	77
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	78 random_state : int or None
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	79 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	80
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	81 def __init__(self, n_iter=1000, positive_thres=-1, negative_thres=0,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	82 verbose=0, n_jobs=1, pre_dispatch='2*n_jobs',
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	83 random_state=None):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	84 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	85 IRAPS turns towwards general Anomaly Detection
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	86 It comapares positive_thres with negative_thres,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	87 and decide which portion is the positive target.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	88 e.g.:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	89 (positive_thres=-1, negative_thres=0)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	90 => positive = Z_score of target < -1
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	91 (positive_thres=1, negative_thres=0)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	92 => positive = Z_score of target > 1
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	93
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	94 Note: The positive targets here is always the
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	95 abnormal minority group.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	96 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	97 self.n_iter = n_iter
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	98 self.positive_thres = positive_thres
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	99 self.negative_thres = negative_thres
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	100 self.verbose = verbose
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	101 self.n_jobs = n_jobs
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	102 self.pre_dispatch = pre_dispatch
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	103 self.random_state = random_state
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	104
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	105 def fit(self, X, y):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	106 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	107 X: array-like (n_samples x n_features)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	108 y: 1-d array-like (n_samples)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	109 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	110 X, y = check_X_y(X, y, ['csr', 'csc'], multi_output=False)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	111
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	112 def _stochastic_sampling(X, y, random_state=None, positive_thres=-1,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	113 negative_thres=0):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	114 # each iteration select a random number of random subset of
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	115 # training samples. this is somewhat different from the original
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	116 # IRAPS method, but effect is almost the same.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	117 SAMPLE_SIZE = [0.25, 0.75]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	118 n_samples = X.shape[0]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	119
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	120 if random_state is None:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	121 n_select = random.randint(int(n_samples * SAMPLE_SIZE[0]),
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	122 int(n_samples * SAMPLE_SIZE[1]))
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	123 index = random.sample(list(range(n_samples)), n_select)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	124 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	125 n_select = random.Random(random_state).randint(
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	126 int(n_samples * SAMPLE_SIZE[0]),
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	127 int(n_samples * SAMPLE_SIZE[1]))
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	128 index = random.Random(random_state).sample(
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	129 list(range(n_samples)), n_select)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	130
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	131 X_selected, y_selected = X[index], y[index]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	132
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	133 # Spliting by z_scores.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	134 y_selected = (y_selected - y_selected.mean()) / y_selected.std()
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	135 if positive_thres < negative_thres:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	136 X_selected_positive = X_selected[y_selected < positive_thres]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	137 X_selected_negative = X_selected[y_selected > negative_thres]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	138 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	139 X_selected_positive = X_selected[y_selected > positive_thres]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	140 X_selected_negative = X_selected[y_selected < negative_thres]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	141
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	142 # For every iteration, at least 5 responders are selected
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	143 if X_selected_positive.shape[0] < 5:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	144 warnings.warn("Warning: fewer than 5 positives were selected!")
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	145 return
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	146
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	147 # p_values
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	148 _, p = ttest_ind(X_selected_positive, X_selected_negative,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	149 axis=0, equal_var=False)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	150
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	151 # fold_change == mean change?
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	152 # TODO implement other normalization method
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	153 positive_mean = X_selected_positive.mean(axis=0)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	154 negative_mean = X_selected_negative.mean(axis=0)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	155 mean_change = positive_mean - negative_mean
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	156 # mean_change = np.select(
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	157 # [positive_mean > negative_mean,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	158 # positive_mean < negative_mean],
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	159 # [positive_mean / negative_mean,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	160 # -negative_mean / positive_mean])
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	161 # mean_change could be adjusted by power of 2
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	162 # mean_change = 2**mean_change \
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	163 # if mean_change>0 else -2**abs(mean_change)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	164
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	165 return p, mean_change, negative_mean
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	166
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	167 parallel = Parallel(n_jobs=self.n_jobs, verbose=self.verbose,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	168 pre_dispatch=self.pre_dispatch)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	169 if self.random_state is None:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	170 res = parallel(delayed(_stochastic_sampling)(
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	171 X, y, random_state=None,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	172 positive_thres=self.positive_thres,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	173 negative_thres=self.negative_thres)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	174 for i in range(self.n_iter))
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	175 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	176 res = parallel(delayed(_stochastic_sampling)(
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	177 X, y, random_state=seed,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	178 positive_thres=self.positive_thres,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	179 negative_thres=self.negative_thres)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	180 for seed in range(self.random_state,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	181 self.random_state+self.n_iter))
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	182 res = [_ for _ in res if _]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	183 if len(res) < 50:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	184 raise ValueError("too few (%d) valid feature lists "
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	185 "were generated!" % len(res))
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	186 pvalues = np.vstack([x[0] for x in res])
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	187 fold_changes = np.vstack([x[1] for x in res])
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	188 base_values = np.vstack([x[2] for x in res])
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	189
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	190 self.pvalues_ = np.asarray(pvalues)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	191 self.fold_changes_ = np.asarray(fold_changes)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	192 self.base_values_ = np.asarray(base_values)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	193
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	194 return self
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	195
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	196
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	197 def _iraps_core_fit(iraps_core, X, y):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	198 return iraps_core.fit(X, y)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	199
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	200
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	201 class IRAPSClassifier(six.with_metaclass(ABCMeta, _BaseFilter,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	202 BaseEstimator, RegressorMixin)):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	203 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	204 Extend the bases of both sklearn feature_selector and classifier.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	205 From sklearn BaseEstimator:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	206 get_params()
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	207 set_params()
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	208 From sklearn _BaseFilter:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	209 get_support()
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	210 fit_transform(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	211 transform(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	212 From sklearn RegressorMixin:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	213 score(X, y): R2
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	214 New:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	215 predict(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	216 predict_label(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	217 get_signature()
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	218 Properties:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	219 discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	220
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	221 Parameters
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	222 ----------
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	223 iraps_core: object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	224 p_thres: float, threshold for p_values
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	225 fc_thres: float, threshold for fold change or mean difference
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	226 occurrence: float, occurrence rate selected by set of p_thres and fc_thres
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	227 discretize: float, threshold of z_score to discretize target value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	228 memory: None, str or joblib.Memory object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	229 min_signature_features: int, the mininum number of features in a signature
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	230 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	231
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	232 def __init__(self, iraps_core, p_thres=1e-4, fc_thres=0.1,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	233 occurrence=0.8, discretize=-1, memory=None,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	234 min_signature_features=1):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	235 self.iraps_core = iraps_core
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	236 self.p_thres = p_thres
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	237 self.fc_thres = fc_thres
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	238 self.occurrence = occurrence
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	239 self.discretize = discretize
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	240 self.memory = memory
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	241 self.min_signature_features = min_signature_features
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	242
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	243 def fit(self, X, y):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	244 memory = check_memory(self.memory)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	245 cached_fit = memory.cache(_iraps_core_fit)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	246 iraps_core = clone(self.iraps_core)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	247 # allow pre-fitted iraps_core here
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	248 if not hasattr(iraps_core, 'pvalues_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	249 iraps_core = cached_fit(iraps_core, X, y)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	250 self.iraps_core_ = iraps_core
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	251
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	252 pvalues = as_float_array(iraps_core.pvalues_, copy=True)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	253 # why np.nan is here?
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	254 pvalues[np.isnan(pvalues)] = np.finfo(pvalues.dtype).max
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	255
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	256 fold_changes = as_float_array(iraps_core.fold_changes_, copy=True)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	257 fold_changes[np.isnan(fold_changes)] = 0.0
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	258
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	259 base_values = as_float_array(iraps_core.base_values_, copy=True)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	260
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	261 p_thres = self.p_thres
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	262 fc_thres = self.fc_thres
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	263 occurrence = self.occurrence
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	264
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	265 mask_0 = np.zeros(pvalues.shape, dtype=np.int32)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	266 # mark p_values less than the threashold
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	267 mask_0[pvalues <= p_thres] = 1
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	268 # mark fold_changes only when greater than the threashold
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	269 mask_0[abs(fold_changes) < fc_thres] = 0
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	270
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	271 # count the occurrence and mask greater than the threshold
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	272 counts = mask_0.sum(axis=0)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	273 occurrence_thres = int(occurrence * iraps_core.n_iter)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	274 mask = np.zeros(counts.shape, dtype=bool)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	275 mask[counts >= occurrence_thres] = 1
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	276
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	277 # generate signature
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	278 fold_changes[mask_0 == 0] = 0.0
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	279 signature = fold_changes[:, mask].sum(axis=0) / counts[mask]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	280 signature = np.vstack((signature, base_values[:, mask].mean(axis=0)))
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	281 # It's not clearn whether min_size could impact prediction
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	282 # performance
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	283 if signature is None\
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	284 or signature.shape[1] < self.min_signature_features:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	285 raise ValueError("The classifier got None signature or the number "
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	286 "of sinature feature is less than minimum!")
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	287
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	288 self.signature_ = np.asarray(signature)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	289 self.mask_ = mask
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	290 # TODO: support other discretize method: fixed value, upper
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	291 # third quater, etc.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	292 self.discretize_value = y.mean() + y.std() * self.discretize
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	293 if iraps_core.negative_thres > iraps_core.positive_thres:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	294 self.less_is_positive = True
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	295 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	296 self.less_is_positive = False
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	297
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	298 return self
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	299
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	300 def _get_support_mask(self):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	301 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	302 return mask of feature selection indices
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	303 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	304 check_is_fitted(self, 'mask_')
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	305
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	306 return self.mask_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	307
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	308 def get_signature(self):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	309 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	310 return signature
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	311 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	312 check_is_fitted(self, 'signature_')
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	313
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	314 return self.signature_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	315
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	316 def predict(self, X):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	317 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	318 compute the correlation coefficient with irpas signature
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	319 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	320 signature = self.get_signature()
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	321
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	322 X = as_float_array(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	323 X_transformed = self.transform(X) - signature[1]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	324 corrcoef = np.array(
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	325 [np.corrcoef(signature[0], e)[0][1] for e in X_transformed])
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	326 corrcoef[np.isnan(corrcoef)] = np.finfo(np.float32).min
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	327
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	328 return corrcoef
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	329
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	330 def predict_label(self, X, clf_cutoff=0.4):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	331 return self.predict(X) >= clf_cutoff
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	332
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	333
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	334 class BinarizeTargetClassifier(BaseEstimator, RegressorMixin):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	335 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	336 Convert continuous target to binary labels (True and False)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	337 and apply a classification estimator.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	338
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	339 Parameters
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	340 ----------
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	341 classifier: object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	342 Estimator object such as derived from sklearn `ClassifierMixin`.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	343
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	344 z_score: float, default=-1.0
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	345 Threshold value based on z_score. Will be ignored when
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	346 fixed_value is set
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	347
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	348 value: float, default=None
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	349 Threshold value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	350
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	351 less_is_positive: boolean, default=True
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	352 When target is less the threshold value, it will be converted
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	353 to True, False otherwise.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	354
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	355 Attributes
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	356 ----------
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	357 classifier_: object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	358 Fitted classifier
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	359
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	360 discretize_value: float
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	361 The threshold value used to discretize True and False targets
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	362 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	363
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	364 def __init__(self, classifier, z_score=-1, value=None,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	365 less_is_positive=True):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	366 self.classifier = classifier
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	367 self.z_score = z_score
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	368 self.value = value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	369 self.less_is_positive = less_is_positive
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	370
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	371 def fit(self, X, y, sample_weight=None):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	372 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	373 Convert y to True and False labels and then fit the classifier
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	374 with X and new y
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	375
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	376 Returns
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	377 ------
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	378 self: object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	379 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	380 y = check_array(y, accept_sparse=False, force_all_finite=True,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	381 ensure_2d=False, dtype='numeric')
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	382 y = column_or_1d(y)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	383
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	384 if self.value is None:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	385 discretize_value = y.mean() + y.std() * self.z_score
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	386 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	387 discretize_value = self.Value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	388 self.discretize_value = discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	389
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	390 if self.less_is_positive:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	391 y_trans = y < discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	392 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	393 y_trans = y > discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	394
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	395 self.classifier_ = clone(self.classifier)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	396
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	397 if sample_weight is not None:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	398 self.classifier_.fit(X, y_trans, sample_weight=sample_weight)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	399 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	400 self.classifier_.fit(X, y_trans)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	401
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	402 if hasattr(self.classifier_, 'feature_importances_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	403 self.feature_importances_ = self.classifier_.feature_importances_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	404 if hasattr(self.classifier_, 'coef_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	405 self.coef_ = self.classifier_.coef_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	406 if hasattr(self.classifier_, 'n_outputs_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	407 self.n_outputs_ = self.classifier_.n_outputs_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	408 if hasattr(self.classifier_, 'n_features_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	409 self.n_features_ = self.classifier_.n_features_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	410
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	411 return self
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	412
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	413 def predict(self, X):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	414 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	415 Predict class probabilities of X.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	416 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	417 check_is_fitted(self, 'classifier_')
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	418 proba = self.classifier_.predict_proba(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	419 return proba[:, 1]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	420
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	421 def predict_label(self, X):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	422 """Predict class label of X
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	423 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	424 check_is_fitted(self, 'classifier_')
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	425 return self.classifier_.predict(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	426
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	427
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	428 class _BinarizeTargetProbaScorer(_BaseScorer):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	429 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	430 base class to make binarized target specific scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	431 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	432
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	433 def __call__(self, clf, X, y, sample_weight=None):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	434 clf_name = clf.__class__.__name__
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	435 # support pipeline object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	436 if isinstance(clf, Pipeline):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	437 main_estimator = clf.steps[-1][-1]
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	438 # support stacking ensemble estimators
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	439 # TODO support nested pipeline/stacking estimators
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	440 elif clf_name in ['StackingCVClassifier', 'StackingClassifier']:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	441 main_estimator = clf.meta_clf_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	442 elif clf_name in ['StackingCVRegressor', 'StackingRegressor']:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	443 main_estimator = clf.meta_regr_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	444 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	445 main_estimator = clf
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	446
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	447 discretize_value = main_estimator.discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	448 less_is_positive = main_estimator.less_is_positive
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	449
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	450 if less_is_positive:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	451 y_trans = y < discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	452 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	453 y_trans = y > discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	454
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	455 y_pred = clf.predict(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	456 if sample_weight is not None:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	457 return self._sign * self._score_func(y_trans, y_pred,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	458 sample_weight=sample_weight,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	459 **self._kwargs)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	460 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	461 return self._sign * self._score_func(y_trans, y_pred,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	462 **self._kwargs)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	463
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	464
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	465 # roc_auc
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	466 binarize_auc_scorer =\
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	467 _BinarizeTargetProbaScorer(metrics.roc_auc_score, 1, {})
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	468
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	469 # average_precision_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	470 binarize_average_precision_scorer =\
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	471 _BinarizeTargetProbaScorer(metrics.average_precision_score, 1, {})
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	472
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	473 # roc_auc_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	474 iraps_auc_scorer = binarize_auc_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	475
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	476 # average_precision_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	477 iraps_average_precision_scorer = binarize_average_precision_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	478
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	479
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	480 class BinarizeTargetRegressor(BaseEstimator, RegressorMixin):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	481 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	482 Extend regression estimator to have discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	483
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	484 Parameters
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	485 ----------
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	486 regressor: object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	487 Estimator object such as derived from sklearn `RegressionMixin`.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	488
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	489 z_score: float, default=-1.0
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	490 Threshold value based on z_score. Will be ignored when
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	491 fixed_value is set
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	492
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	493 value: float, default=None
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	494 Threshold value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	495
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	496 less_is_positive: boolean, default=True
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	497 When target is less the threshold value, it will be converted
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	498 to True, False otherwise.
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	499
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	500 Attributes
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	501 ----------
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	502 regressor_: object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	503 Fitted regressor
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	504
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	505 discretize_value: float
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	506 The threshold value used to discretize True and False targets
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	507 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	508
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	509 def __init__(self, regressor, z_score=-1, value=None,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	510 less_is_positive=True):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	511 self.regressor = regressor
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	512 self.z_score = z_score
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	513 self.value = value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	514 self.less_is_positive = less_is_positive
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	515
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	516 def fit(self, X, y, sample_weight=None):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	517 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	518 Calculate the discretize_value fit the regressor with traning data
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	519
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	520 Returns
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	521 ------
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	522 self: object
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	523 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	524 y = check_array(y, accept_sparse=False, force_all_finite=True,
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	525 ensure_2d=False, dtype='numeric')
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	526 y = column_or_1d(y)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	527
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	528 if self.value is None:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	529 discretize_value = y.mean() + y.std() * self.z_score
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	530 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	531 discretize_value = self.Value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	532 self.discretize_value = discretize_value
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	533
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	534 self.regressor_ = clone(self.regressor)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	535
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	536 if sample_weight is not None:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	537 self.regressor_.fit(X, y, sample_weight=sample_weight)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	538 else:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	539 self.regressor_.fit(X, y)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	540
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	541 # attach classifier attributes
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	542 if hasattr(self.regressor_, 'feature_importances_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	543 self.feature_importances_ = self.regressor_.feature_importances_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	544 if hasattr(self.regressor_, 'coef_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	545 self.coef_ = self.regressor_.coef_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	546 if hasattr(self.regressor_, 'n_outputs_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	547 self.n_outputs_ = self.regressor_.n_outputs_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	548 if hasattr(self.regressor_, 'n_features_'):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	549 self.n_features_ = self.regressor_.n_features_
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	550
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	551 return self
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	552
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	553 def predict(self, X):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	554 """Predict target value of X
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	555 """
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	556 check_is_fitted(self, 'regressor_')
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	557 y_pred = self.regressor_.predict(X)
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	558 if not np.all((y_pred >= 0) & (y_pred <= 1)):
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	559 y_pred = (y_pred - y_pred.min()) / (y_pred.max() - y_pred.min())
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	560 if self.less_is_positive:
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	561 y_pred = 1 - y_pred
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	562 return y_pred
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	563
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	564
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	565 # roc_auc_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	566 regression_auc_scorer = binarize_auc_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	567
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	568 # average_precision_scorer
1c4a241bef5c planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit c0a3a186966888e5787335a7628bf0a4382637e7 bgruening parents: diff changeset	569 regression_average_precision_scorer = binarize_average_precision_scorer

Mercurial > repos > bgruening > sklearn_searchcv

annotate iraps_classifier.py @ 8:1c4a241bef5c draft