-
Notifications
You must be signed in to change notification settings - Fork 0
/
active_learning.py
399 lines (383 loc) · 14.5 KB
/
active_learning.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
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
from sklearn.naive_bayes import MultinomialNB
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.feature_extraction.text import HashingVectorizer
from sklearn import metrics
import pandas as pd
import unicodedata
import numpy as np
from sklearn.svm import SVC
from compiler.ast import flatten
from openpyxl import Workbook, load_workbook
from scipy.stats.stats import pearsonr
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier,GradientBoostingClassifier
from sklearn.linear_model import SGDClassifier
import nltk
import nltk as nl
from nltk import re
from nltk.tokenize import word_tokenize
from nltk.tokenize import sent_tokenize
from nltk.chunk.regexp import RegexpParser
from nltk.tag.util import *
from nltk.corpus import stopwords
from sklearn.metrics import confusion_matrix
import os
import re
#from tsne import bh_sne
import matplotlib.pyplot as plt
import pylab
from sklearn.metrics import roc_auc_score,f1_score
from sklearn.metrics import roc_curve, auc
from sklearn.cross_validation import train_test_split
from sklearn.preprocessing import label_binarize
from sklearn.multiclass import OneVsRestClassifier
from sklearn.metrics import roc_curve, auc
import math as m
from sklearn.preprocessing import normalize
from sklearn.cross_validation import cross_val_score, KFold
from sklearn.cross_validation import train_test_split
from datetime import datetime
start_time = datetime.now()
label_triple = []
def entropy(grad_label0,grad_label1):
ME = -(( grad_label0 * m.log(grad_label0,2)) + grad_label1 * m.log(grad_label1,2))
return ME
################################single random permutation checking#################################################################
df = pd.read_excel('/home/mohan/Downloads/Theano-Tutorials-master/vnsp_write_features.xlsx',sheetname=1)
#data = df.iloc[:,1:5]
train_test = df.iloc[:,[1,2,3,4,5,6]]
for j in range(np.shape(train_test)[1]):
for i in range(np.shape(train_test)[0]):
string = train_test.iloc[i,j]
if(type(string) == float):
if(str(float(string)).lower() == 'nan'):
train_test.iloc[i,j] = df.iloc[i,1]
data = np.asarray(train_test)
#data = np.random.permutation(data)
train_data,test_data = train_test_split(data , test_size=0.05, random_state=42)
train_check = train_data
#print "the type of train_data is"
#print type(train_data)
#print "the indices of the train_data"
#print np.unique(train_data, return_index= True)[0]
#print "the type of test_data is"
#print type(test_data)
#print train_check[0,0]
#print "type of original data"
#print type(train_data)
#print "original train_data"
#print train_data[0,0]
#print "original label"
#print train_data[0,1]
cv = KFold(n=train_data.shape[0], # total number of samples
n_folds=5, # number of folds the dataset is divided into
random_state=12345)
#print "length of cv"
#print cv
#print '#########################test_data########################################################'
#print '##########################################################################################'
for i in range(len(train_data)):
if(train_data[i,1]=='VNSP'):
train_data[i,1] = 1
else:
train_data[i,1] = 0
label_train = train_data[:,1]
for i in range(len(test_data)):
if(test_data[i,1]=='VNSP'):
test_data[i,1] = 1
else:
test_data[i,1] = 0
label_test = test_data[:,1]
vctr = CountVectorizer(stop_words='english',min_df = 1)
vctr2 = HashingVectorizer(stop_words='english')
vctr1 = TfidfVectorizer(stop_words='english')
count_pos = 0
count_neg = 0
######################################################################################################
train = []
test = []
for i in range(len(train_data)):
string = train_data[i,0]
string = vctr.build_preprocessor()(string.lower())
string = vctr.build_tokenizer()(string.lower())
train.append(' '.join(string))
for i in range(len(test_data)):
string = test_data[i,0]
string = vctr.build_preprocessor()(string.lower())
string = vctr.build_tokenizer()(string.lower())
test.append(' '.join(string))
######################################################################################################
train_data1 = vctr.fit_transform(train).toarray()
#X_test = vctr.transform(test).toarray()
y_train = np.asarray(label_train, dtype="|S6")
y_train = y_train.astype(int)
clf1 = GradientBoostingClassifier(n_estimators = 100)
clf2 = AdaBoostClassifier(n_estimators = 100)
clf3 = RandomForestClassifier(n_estimators = 100)
#print "type of train_data"
#print type(train_data1)
#print "type of y_train"
#print type(y_train)
#print "type of cv"
#print type(cv)
scores1 = cross_val_score(clf1, train_data1 , y_train, cv=cv, scoring='mean_absolute_error')
scores2 = cross_val_score(clf2, train_data1 , y_train, cv=cv, scoring='mean_absolute_error')
scores3 = cross_val_score(clf3, train_data1 , y_train, cv=cv, scoring='mean_absolute_error')
scores4 = cross_val_score(clf1, train_data1 , y_train, cv=cv, scoring='f1')
scores5 = cross_val_score(clf2, train_data1 , y_train, cv=cv, scoring='f1')
scores6 = cross_val_score(clf3, train_data1 , y_train, cv=cv, scoring='f1')
score1=[]
score2=[]
score3=[]
score4=[]
score5=[]
score6=[]
val_3classifiers = {}
for score in scores1:
#print "score"
#print score
score1.append(score)
for score in scores4:
#print "score"
#print score
score4.append(score)
#print "mean error for classifier 1 is"
#print np.mean(score1)
#print "minimum error for classifier 1 is"
#print max(score1)
val1, idx1 = max((val1, idx1) for (idx1, val1) in enumerate(score1))
#print "minimum error and index for classifier1"
#print val1
#print idx1
###################################################################################################################################
#print "mean f1_score of classifier1"
#print np.mean(score4)
#print "max f1 and its index for classifier1"
valf1, idxf1 = max((valf1, idxf1) for (idxf1, valf1) in enumerate(score4))
#print valf1
#print idxf1
########################################################################################################################################
for score in scores2:
#print "score"
#print score
score2.append(score)
for score in scores5:
#print "score"
#print score
score5.append(score)
#print "mean error for classifier 2 is"
#print np.mean(score2)
#print "minimum error for classifier 2 is"
#print max(score2)
val2, idx2 = max((val2, idx2) for (idx2, val2) in enumerate(score2))
#print "minimum error and index for classifier2"
#print val2
#print idx2
###################################################################################################################################
#print "mean f1_score of classifier2"
#print np.mean(score5)
#print "max f2 and its index for classifier2"
valf2, idxf2 = max((valf2, idxf2) for (idxf2, valf2) in enumerate(score5))
#print valf2
#print idxf2
###########################################################################################################################################
for score in scores3:
#print "score"
#print score
score3.append(score)
for score in scores6:
#print "score"
#print score
score6.append(score)
#print "mean error for classifier 3 is"
#print np.mean(score3)
#print "minimum error for classifier 3 is"
#print max(score3)
val3, idx3 = max((val3, idx3) for (idx3, val3) in enumerate(score3))
#print "minimum error and index for classifier3"
#print val3
#print idx3
###################################################################################################################################
#print "mean f1_score of classifier3"
#print np.mean(score6)
#print "max f3 and its index for classifier3"
valf3, idxf3 = max((valf3, idxf3) for (idxf3, valf3) in enumerate(score6))
#print valf3
#print idxf3
###########################################################################################################################################
val_class = {}
val_class[val1]=idx1
val_class[val2]=idx2
val_class[val3]=idx3
#print "the minimum value of class and its index is"
for val in val_class:
if(val == max(val_class.keys())):
k = val
print k,val_class[k]
print "the global minimum error"
print k
print "the global minimum index"
print val_class[k]
required_train_data = []
required_test_data = []
count = 0
for train, test1 in cv:
#print "normal indices are"
#print train
#print test
if(count == val_class[k]):
#print "a"
#print "required indices"
#print("%s %s" % (train, test))
#print "required data set"
#print "length of the train data and its data set"
#print train_check[train,0]
#print "label of data is"
#print train_check[train,1]
required_train_data.append(train)
required_train_data.append(test1)
count = count + 1
#print " the count value is"
#print count
#print "the concate"
required_train_data = np.concatenate(required_train_data)
#print "length of requored_train_data "
#print len(required_train_data)
#print required_train_data[0]
train_data[required_train_data,0]
label_train1 = []
train_check=[]
for i in range(len(required_train_data)):
#print "a"
string = train_data[required_train_data[i],0]
#print string
label_train1.append(train_data[required_train_data[i],1])
string = vctr1.build_preprocessor()(string.lower())
string = vctr1.build_tokenizer()(string.lower())
train_check.append(' '.join(string))
train_data2 = vctr.fit_transform(train_check).toarray()
#print "X_test"
#print X_test
y_train1 = np.asarray(label_train1, dtype="|S6")
y_train1 = y_train1.astype(int)
test_data_labels=y_train1
#print "real labels"
#print y_train1
#clf1 = GradientBoostingClassifier(n_estimators = 1000)
#clf2 = AdaBoostClassifier(n_estimators = 1000)
#clf3 = RandomForestClassifier()
#print "the test data is"
#print test[1:3]
f=open('/home/mohan/Downloads/Theano-Tutorials-master/crawledcontents95.txt')
raw = f.read()
u1 = unicode(raw, "utf-8")
string = unicodedata.normalize('NFKD', u1).encode('ascii','ignore')
sent = nl.sent_tokenize(string)
test_sents = []
for i in range(len(sent)):
test_sents.append(' '.join(nl.word_tokenize(re.sub('''[!@#$(),.;'":/\n?!\W]''',' ', sent[i]))))
###########################################################################################################################################
clf1.fit(train_data2,y_train1)
X_test1 = vctr.transform(test_sents).toarray()
grad_label = clf1.predict(X_test1)
grad_label = grad_label.astype(int)
print "grad_label"
print grad_label
for i in range(len(grad_label)):
print test_sents[i] ,'=>', grad_label[i]
################################################grad prob#######################################
clf1.fit(train_data2,y_train1)
X_test1 = vctr.transform(test_sents).toarray()
grad_label1 = clf1.predict_proba(X_test1)
#grad_label1 = grad_label.astype(int)
print "grad_label with probablistic labels "
print grad_label1
for i in range(len(grad_label1)):
print test_sents[i] ,'=>', grad_label1[i]
##########################################################################################
clf2.fit(train_data2,y_train1)
X_test2 = vctr.transform(test_sents).toarray()
ada_label = clf2.predict(X_test2)
ada_label = ada_label.astype(int)
print "ada_label"
print ada_label
for i in range(len(ada_label)):
print test_sents[i] ,'=>', ada_label[i]
###########################################################################################
clf2.fit(train_data2,y_train1)
X_test2 = vctr.transform(test_sents).toarray()
ada_label1= clf2.predict_proba(X_test2)
#ada_label = ada_label.astype(int)
print "ada_label with probabalistic labels"
for i in range(len(ada_label1)):
print test_sents[i] ,'=>', ada_label1[i]
###########################3
clf3.fit(train_data2,y_train1)
X_test3 = vctr.transform(test_sents).toarray()
rand_label = clf3.predict(X_test3)
rand_label = rand_label.astype(int)
print "rand_label of true labels"
print rand_label
for i in range(len(rand_label)):
print test_sents[i] ,'=>', rand_label[i]
labelgot = []
for i in range(len(rand_label)):
avg = float(float(grad_label[i]+ada_label[i]+grad_label[i])/(float(3)))
labelgot.append(avg)
print "labels averaged"
print labelgot
for i in range(len(labelgot)):
if(labelgot[i] == 0.3333333333333333):
labelgot[i] = 0.0
elif(labelgot[i] == 0.6666666666666666):
labelgot[i] = 1.0
elif(labelgot[i] == 1):
labelgot[i] = 1.0
elif(labelgot[i] == 0):
labelgot[i] = 0.0
print "############################################################ developed label set ##################################################"
print "developed label set type "
#print labelgot
print "#############################################################################################"
print
print "active learning"
print
print "##############################################################################################"
c = 0
for i in range(len(rand_label)):
if(rand_label[i] == grad_label[i] == ada_label[i]):
c = c + 1
print "can be accepted as a label"+' '+str(i)+' '+str(test_sents[i])
elif(rand_label[i] != grad_label[i] == ada_label[i]):
print "can be accepted as a label"+' '+str(i)+' '+str(test_sents[i])
elif(rand_label[i] == ada_label[i] != grad_label[i]):
print "can be accepted as a label"+' '+str(i)+' '+str(test_sents[i])
#elif(rand_label[i] != grad_label[i] == ada_label[i]):
# print "to be queried by the committe"
# print str(i)+str(=>)+test_sents[i]
else:
print "k"
print "to be queried by the committe"+ ' '+str(i)+' '+str(test_sents[i])
print "not entered the loop"
print c
print len(rand_label)
#print "rand_label"
#print rand_label
entropy_outputs1=[]
entropy_outputs2=[]
##################################################checking entropy of the labels#######################
for i in range(len(grad_label1)):
print grad_label1[i],'=>',entropy(grad_label1[i][0],grad_label1[i][1])
entropy_outputs1.append(entropy(grad_label1[i][0],grad_label1[i][1]))
print "entropy for grad_label"
print entropy_outputs1
for i in range(len(ada_label1)):
print ada_label1[i],'=>',entropy(ada_label1[i][0],ada_label1[i][1])
entropy_outputs2.append(entropy(ada_label1[i][0],ada_label1[i][1]))
print "entropy for ada_label"
print entropy_outputs2
#required_labels = increasing_measure(grad_label,ada_label,rand_label)
#print "required meta voting labels"
end_time = datetime.now()
print('Duration: {}'.format(end_time - start_time))