-
Notifications
You must be signed in to change notification settings - Fork 20
/
estimator.py
303 lines (269 loc) · 11.6 KB
/
estimator.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
"""
Cause-effect models.
"""
# Author: Jose A. R. Fonollosa <jarfo@yahoo.com>
#
# License: Apache, Version 2.0
import features as f
import numpy as np
from sklearn import pipeline
from sklearn.base import BaseEstimator
from sklearn.ensemble import GradientBoostingClassifier
from multiprocessing import Pool
gbc_params = {
'loss':'deviance',
'learning_rate': 0.1,
'n_estimators': 500,
'subsample': 1.0,
'min_samples_split': 8,
'min_samples_leaf': 1,
'max_depth': 9,
'init': None,
'random_state': 1,
'max_features': None,
'verbose': 0
}
selected_features = [
'Adjusted Mutual Information[A,A type,B,B type]',
'Conditional Distribution Entropy Variance[A,A type,B,B type]',
'Conditional Distribution Entropy Variance[B,B type,A,A type]',
'Conditional Distribution Kurtosis Variance[A,A type,B,B type]',
'Conditional Distribution Kurtosis Variance[B,B type,A,A type]',
'Conditional Distribution Similarity[A,A type,B,B type]',
'Conditional Distribution Similarity[B,B type,A,A type]',
'Conditional Distribution Skewness Variance[A,A type,B,B type]',
'Conditional Distribution Skewness Variance[B,B type,A,A type]',
'Discrete Conditional Entropy[A,A type,B,B type]',
'Discrete Conditional Entropy[B,B type,A,A type]',
'Discrete Entropy[A,A type]',
'Discrete Entropy[B,B type]',
'Discrete Mutual Information[A,A type,B,B type]',
'HSIC[A,A type,B,B type]',
'IGCI[A,A type,B,B type]',
'IGCI[B,B type,A,A type]',
'Kurtosis[A,A type]',
'Kurtosis[B,B type]',
'Log[Number of Samples[A]]',
'Log[Number of Unique Samples[A]]',
'Log[Number of Unique Samples[B]]',
'Moment21[A,A type,B,B type]',
'Moment21[B,B type,A,A type]',
'Moment31[A,A type,B,B type]',
'Moment31[B,B type,A,A type]',
'Normalized Discrete Entropy[A,A type]',
'Normalized Discrete Entropy[B,B type]',
'Normalized Discrete Mutual Information[Discrete Mutual Information[A,A type,B,B type],Discrete Joint Entropy[A,A type,B,B type]]',
'Normalized Discrete Mutual Information[Discrete Mutual Information[A,A type,B,B type],Min[Discrete Entropy[A,A type],Discrete Entropy[B,B type]]]',
'Normalized Entropy[A,A type]',
'Normalized Entropy[B,B type]',
'Normalized Error Probability[A,A type,B,B type]',
'Normalized Error Probability[B,B type,A,A type]',
# 'Number of Unique Samples[A]',
# 'Number of Unique Samples[B]',
'Pearson R[A,A type,B,B type]',
'Polyfit Error[A,A type,B,B type]',
'Polyfit Error[B,B type,A,A type]',
'Polyfit[A,A type,B,B type]',
'Polyfit[B,B type,A,A type]',
'Skewness[A,A type]',
'Skewness[B,B type]',
'Uniform Divergence[A,A type]',
'Uniform Divergence[B,B type]'
]
class Pipeline(pipeline.Pipeline):
def predict(self, X):
try:
p = super(Pipeline, self).predict_proba(X)
if p.shape[1] == 2:
p = p[:,1]
elif p.shape[1] == 3:
p = p[:,2] - p[:,0]
except AttributeError:
p = super(Pipeline, self).predict(X)
return p
def get_pipeline(features, regressor=None, params=None):
steps = [
("extract_features", f.FeatureMapper(features)),
("regressor", regressor(**params)),
]
return Pipeline(steps)
class CauseEffectEstimatorOneStep(BaseEstimator):
def __init__(self, features=None, regressor=None, params=None, symmetrize=True):
self.extractor = f.extract_features
self.classifier = get_pipeline(features, regressor, params)
self.symmetrize = symmetrize
def extract(self, features):
return self.extractor(features)
def fit(self, X, y=None):
self.classifier.fit(X, y)
return self
def fit_transform(self, X, y=None):
return self.classifier.fit_transform(X, y)
def transform(self, X):
return self.classifier.transform(X)
def predict(self, X):
predictions = self.classifier.predict(X)
if self.symmetrize:
predictions[0::2] = (predictions[0::2] - predictions[1::2])/2
predictions[1::2] = -predictions[0::2]
return predictions
class CauseEffectEstimatorSymmetric(BaseEstimator):
def __init__(self, features=None, regressor=None, params=None, symmetrize=True):
self.extractor = f.extract_features
self.classifier_left = get_pipeline(features, regressor, params)
self.classifier_right = get_pipeline(features, regressor, params)
self.symmetrize = symmetrize
def extract(self, features):
return self.extractor(features)
def fit(self, X, y=None):
target_left = np.array(y)
target_left[target_left != 1] = 0
weight_left = np.ones(len(target_left))
weight_left[target_left==0] = sum(target_left==1)/float(sum(target_left==0))
try:
self.classifier_left.fit(X, target_left, regressor__sample_weight=weight_left)
except TypeError:
self.classifier_left.fit(X, target_left)
target_right = np.array(y)
target_right[target_right != -1] = 0
target_right[target_right == -1] = 1
weight_right = np.ones(len(target_right))
weight_right[target_right==0] = sum(target_right==1)/float(sum(target_right==0))
try:
self.classifier_right.fit(X, target_right, regressor__sample_weight=weight_right)
except TypeError:
self.classifier_right.fit(X, target_right)
return self
def fit_transform(self, X, y=None):
target_left = np.array(y)
target_left[target_left != 1] = 0
X_left = self.classifier_left.fit_transform(X, target_left)
target_right = np.array(y)
target_right[target_right != -1] = 0
target_right[target_right == -1] = 1
X_right = self.classifier_right.fit_transform(X, target_right)
return X_left, X_right
def transform(self, X):
return self.classifier_left.transform(X), self.classifier_right.transform(X)
def predict(self, X):
predictions_left = self.classifier_left.predict(X)
predictions_right = self.classifier_right.predict(X)
predictions = predictions_left - predictions_right
if self.symmetrize:
predictions[0::2] = (predictions[0::2] - predictions[1::2])/2
predictions[1::2] = -predictions[0::2]
return predictions
class CauseEffectEstimatorID(BaseEstimator):
def __init__(self, features_independence=None, features_direction=None, regressor=None, params=None, symmetrize=True):
self.extractor = f.extract_features
self.classifier_independence = get_pipeline(features_independence, regressor, params)
self.classifier_direction = get_pipeline(features_direction, regressor, params)
self.symmetrize = symmetrize
def extract(self, features):
return self.extractor(features)
def fit(self, X, y=None):
#independence training pairs
train_independence = X
target_independence = np.array(y)
target_independence[target_independence != 0] = 1
weight_independence = np.ones(len(target_independence))
weight_independence[target_independence==0] = sum(target_independence==1)/float(sum(target_independence==0))
try:
self.classifier_independence.fit(train_independence, target_independence, regressor__sample_weight=weight_independence)
except TypeError:
self.classifier_independence.fit(train_independence, target_independence)
#direction training pairs
direction_filter = y != 0
train_direction = X[direction_filter]
target_direction = y[direction_filter]
weight_direction = np.ones(len(target_direction))
weight_direction[target_direction==0] = sum(target_direction==1)/float(sum(target_direction==0))
try:
self.classifier_direction.fit(train_direction, target_direction, regressor__sample_weight=weight_direction)
except TypeError:
self.classifier_direction.fit(train_direction, target_direction)
return self
def fit_transform(self, X, y=None):
#independence training pairs
train_independence = X
target_independence = np.array(y)
target_independence[target_independence != 0] = 1
X_ind = self.classifier_independence.fit_transform(train_independence, target_independence)
#direction training pairs
direction_filter = y != 0
train_direction = X[direction_filter]
target_direction = y[direction_filter]
self.classifier_direction.fit(train_direction, target_direction)
X_dir = self.classifier_direction.transform(X)
return X_ind, X_dir
def transform(self, X):
X_ind = self.classifier_independence.transform(X)
X_dir = self.classifier_direction.transform(X)
return X_ind, X_dir
def predict(self, X):
predictions_independence = self.classifier_independence.predict(X)
if self.symmetrize:
predictions_independence[0::2] = (predictions_independence[0::2] + predictions_independence[1::2])/2
predictions_independence[1::2] = predictions_independence[0::2]
assert predictions_independence.min() >= 0
predictions_direction = self.classifier_direction.predict(X)
if self.symmetrize:
predictions_direction[0::2] = (predictions_direction[0::2] - predictions_direction[1::2])/2
predictions_direction[1::2] = -predictions_direction[0::2]
return predictions_independence * predictions_direction
def calculate_method(args):
obj = args[0]
name = args[1]
margs = args[2]
method = getattr(obj, name)
return method(*margs)
def pmap(func, mlist, n_jobs):
if n_jobs != 1:
pool = Pool(n_jobs if n_jobs != -1 else None)
mmap = pool.map
else:
mmap = map
return mmap(func, mlist)
class CauseEffectSystemCombination(BaseEstimator):
def __init__(self, extractor=f.extract_features, weights=None, symmetrize=True, n_jobs=-1):
self.extractor = extractor
self.features = selected_features
self.systems = [
CauseEffectEstimatorID(
features_direction=self.features,
features_independence=self.features,
regressor=GradientBoostingClassifier,
params=gbc_params,
symmetrize=symmetrize),
CauseEffectEstimatorSymmetric(
features=self.features,
regressor=GradientBoostingClassifier,
params=gbc_params,
symmetrize=symmetrize),
CauseEffectEstimatorOneStep(
features=self.features,
regressor=GradientBoostingClassifier,
params=gbc_params,
symmetrize=symmetrize),
]
self.weights = weights
self.n_jobs = n_jobs
def extract(self, features):
return self.extractor(features, n_jobs=self.n_jobs)
def fit(self, X, y=None):
task = [(m, 'fit', (X, y)) for m in self.systems]
self.systems = pmap(calculate_method, task, self.n_jobs)
return self
def fit_transform(self, X, y=None):
task = [(m, 'fit_transform', (X, y)) for m in self.systems]
return pmap(calculate_method, task, self.n_jobs)
def transform(self, X):
task = [(m, 'transform', (X,)) for m in self.systems]
return pmap(calculate_method, task, self.n_jobs)
def predict(self, X):
task = [(m, 'predict', (X,)) for m in self.systems]
a = np.array(pmap(calculate_method, task, self.n_jobs))
if self.weights is not None:
return np.dot(self.weights, a)
else:
return a