training-crossvalidation-testing-dom-v1.py
15.6 KB
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
# -*- encoding: utf-8 -*-
import os
from time import time
import argparse
import scipy
from sklearn import model_selection
from sklearn.naive_bayes import MultinomialNB, BernoulliNB
from sklearn.svm import SVC
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, confusion_matrix, \
classification_report
from sklearn.externals import joblib
from nltk.corpus import stopwords
from sklearn.feature_extraction.text import TfidfVectorizer, CountVectorizer
from sklearn.feature_selection import SelectKBest, chi2
from sklearn.decomposition import TruncatedSVD
from scipy.sparse import csr_matrix
# Author:
# Carlos Méndez-Cruz
# Goal: training, crossvalidation and testing transcription factor structural domain sentences
# Parameters:
# 1) --inputPath Path to read input files.
# 2) --inputTrainingData File to read training data.
# 4) --inputTrainingClasses File to read training classes.
# 3) --inputTestingData File to read testing data.
# 4) --inputTestingClasses File to read testing classes.
# 5) --outputModelPath Path to place output model.
# 6) --outputModelFile File to place output model.
# 7) --outputReportPath Path to place evaluation report.
# 8) --outputReportFile File to place evaluation report.
# 9) --classifier Classifier: BernoulliNB, SVM, kNN.
# 10) --saveData Save matrices
# 11) --kernel Kernel
# 12) --reduction Feature selection or dimensionality reduction
# 13) --removeStopWords Remove most frequent words
# 14) --ngrinitial N-gram inicial
# 15) --ngrfinal N-gram final
# 14) --vectorizer Vectorizer: b=binary, f=frequency, t=tf-idf.
# Ouput:
# 1) Classification model and evaluation report.
# Execution:
# source activate python3
# python training-crossvalidation-testing-dom-v1.py
# --inputPath /home/text-dom-dataset
# --inputTrainingData train-data.txt
# --inputTrainingClasses train-classes.txt
# --inputTestingData test-data.txt
# --inputTestingClasses test-classes.txt
# --outputModelPath /home/text-dom-dataset/models
# --outputModelFile SVM-lineal-model.mod
# --outputReportPath /home/text-dom-dataset/reports
# --outputReportFile SVM-linear.txt
# --classifier SVM
# --saveData
# --kernel linear
# --vectorizer b
# --ngrinitial 1
# --ngrfinal 1
# python training-crossvalidation-testing-dom-v1.py --inputPath /home/laigen-supervised-learning/text-dom-dataset --inputTrainingData train-data.txt --inputTrainingClasses train-classes.txt --inputTestingData test-data.txt --inputTestingClasses test-classes.txt --outputModelPath /home/laigen-supervised-learning/text-dom-dataset/models --outputModelFile SVM-lineal-model.mod --outputReportPath /home/laigen-supervised-learning/text-dom-dataset/reports --outputReportFile SVM-linear.txt --classifier SVM --saveData --kernel linear --vectorizer b --ngrinitial 1 --ngrfinal 1
###########################################################
# MAIN PROGRAM #
###########################################################
if __name__ == "__main__":
# Parameter definition
parser = argparse.ArgumentParser(description='training, crossvalidation and testing transcription factor structural domain sentences.')
parser.add_argument("--inputPath", dest="inputPath",
help="Path to read input files", metavar="PATH")
parser.add_argument("--inputTrainingData", dest="inputTrainingData",
help="File to read training data", metavar="FILE")
parser.add_argument("--inputTrainingClasses", dest="inputTrainingClasses",
help="File to read training classes", metavar="FILE")
parser.add_argument("--inputTestingData", dest="inputTestingData",
help="File to read testing data", metavar="FILE")
parser.add_argument("--inputTestingClasses", dest="inputTestingClasses",
help="File to read testing classes", metavar="FILE")
parser.add_argument("--outputModelPath", dest="outputModelPath",
help="Path to place output model", metavar="PATH")
parser.add_argument("--outputModelFile", dest="outputModelFile",
help="File to place output model", metavar="FILE")
parser.add_argument("--outputReportPath", dest="outputReportPath",
help="Path to place evaluation report", metavar="PATH")
parser.add_argument("--outputReportFile", dest="outputReportFile",
help="File to place evaluation report", metavar="FILE")
parser.add_argument("--classifier", dest="classifier",
help="Classifier", metavar="NAME",
choices=('BernoulliNB', 'SVM', 'kNN'), default='SVM')
parser.add_argument("--saveData", dest="saveData", action='store_true',
help="Save matrices")
parser.add_argument("--kernel", dest="kernel",
help="Kernel SVM", metavar="NAME",
choices=('linear', 'rbf', 'poly'), default='linear')
parser.add_argument("--reduction", dest="reduction",
help="Feature selection or dimensionality reduction", metavar="NAME",
choices=('SVD200', 'SVD300', 'CHI250', 'CHI2100'), default=None)
parser.add_argument("--removeStopWords", default=False,
action="store_true", dest="removeStopWords",
help="Remove stop words")
parser.add_argument("--ngrinitial", type=int,
dest="ngrinitial", default=1,
help="Initial n-gram", metavar="INTEGER")
parser.add_argument("--ngrfinal", type=int,
dest="ngrfinal", default=1,
help="Final n-gram", metavar="INTEGER")
parser.add_argument("--vectorizer", dest="vectorizer", required=True,
help="Vectorizer: b=binary, f=frequency, t=tf-idf", metavar="CHAR",
choices=('b', 'f', 't'), default='b')
args = parser.parse_args()
# Printing parameter values
print('-------------------------------- PARAMETERS --------------------------------')
print("Path to read input files: " + str(args.inputPath))
print("File to read training data: " + str(args.inputTrainingData))
print("File to read training classes: " + str(args.inputTrainingClasses))
print("File to read testing data: " + str(args.inputTestingData))
print("File to read testing classes: " + str(args.inputTestingClasses))
print("Path to place output model: " + str(args.outputModelPath))
print("File to place output model: " + str(args.outputModelFile))
print("Path to place evaluation report: " + str(args.outputReportPath))
print("File to place evaluation report: " + str(args.outputReportFile))
print("Classifier: " + str(args.classifier))
print("Save matrices: " + str(args.saveData))
print("Kernel: " + str(args.kernel))
print("Reduction: " + str(args.reduction))
print("Remove stop words: " + str(args.removeStopWords))
print("Initial ngram: " + str(args.ngrinitial))
print("Final ngram: " + str(args.ngrfinal))
print("Vectorizer: " + str(args.vectorizer))
# Start time
t0 = time()
if args.removeStopWords:
pf = stopwords.words('english')
else:
pf = None
y_train = []
trainingData = []
y_test = []
testingData = []
X_train = None
X_test = None
if args.saveData:
print("Reading training data and true classes...")
with open(os.path.join(args.inputPath, args.inputTrainingClasses), encoding='utf8', mode='r') \
as iFile:
for line in iFile:
line = line.strip('\r\n')
y_train.append(line)
with open(os.path.join(args.inputPath, args.inputTrainingData), encoding='utf8', mode='r') \
as iFile:
for line in iFile:
line = line.strip('\r\n')
trainingData.append(line)
print(" Done!")
print("Reading testing data and true classes...")
with open(os.path.join(args.inputPath, args.inputTestingClasses), encoding='utf8', mode='r') \
as iFile:
for line in iFile:
line = line.strip('\r\n')
y_test.append(line)
with open(os.path.join(args.inputPath, args.inputTestingData), encoding='utf8', mode='r') \
as iFile:
for line in iFile:
line = line.strip('\r\n')
testingData.append(line)
print(" Done!")
# Create vectorizer
print('Vectorization: {}'.format(args.vectorizer))
if args.vectorizer == "b":
# Binary vectorizer
vectorizer = CountVectorizer(ngram_range=(args.ngrinitial, args.ngrfinal), binary=True, stop_words=pf)
elif args.vectorizer == "f":
# Frequency vectorizer
vectorizer = CountVectorizer(ngram_range=(args.ngrinitial, args.ngrfinal), stop_words=pf)
else:
# Binary vectorizer
vectorizer = TfidfVectorizer(ngram_range=(args.ngrinitial, args.ngrfinal), stop_words=pf)
X_train = csr_matrix(vectorizer.fit_transform(trainingData), dtype='double')
X_test = csr_matrix(vectorizer.transform(testingData), dtype='double')
print(" Saving matrix and classes...")
joblib.dump(X_train, os.path.join(args.outputModelPath, args.inputTrainingData + '.jlb'))
joblib.dump(y_train, os.path.join(args.outputModelPath, args.inputTrainingData + '.class.jlb'))
joblib.dump(X_test, os.path.join(args.outputModelPath, args.inputTestingData + '.jlb'))
joblib.dump(y_test, os.path.join(args.outputModelPath, args.inputTestingClasses + '.class.jlb'))
print(" Done!")
else:
print(" Loading matrix and classes...")
X_train = joblib.load(os.path.join(args.outputModelPath, args.inputTrainingData + '.jlb'))
y_train = joblib.load(os.path.join(args.outputModelPath, args.inputTrainingData + '.class.jlb'))
X_test = joblib.load(os.path.join(args.outputModelPath, args.inputTestingData + '.jlb'))
y_test = joblib.load(os.path.join(args.outputModelPath, args.inputTestingClasses + '.class.jlb'))
print(" Done!")
print(" Number of training classes: {}".format(len(y_train)))
print(" Number of training class DOM: {}".format(y_train.count('DOM')))
print(" Number of training class OTHER: {}".format(y_train.count('OTHER')))
print(" Shape of training matrix: {}".format(X_train.shape))
print(" Number of testing classes: {}".format(len(y_test)))
print(" Number of testing class DOM: {}".format(y_test.count('DOM')))
print(" Number of testing class OTHER: {}".format(y_test.count('OTHER')))
print(" Shape of testing matrix: {}".format(X_test.shape))
# Feature selection and dimensional reduction
if args.reduction is not None:
print('Performing dimensionality reduction or feature selection...', args.reduction)
if args.reduction == 'SVD200':
reduc = TruncatedSVD(n_components=200, random_state=42)
X_train = reduc.fit_transform(X_train)
if args.reduction == 'SVD300':
reduc = TruncatedSVD(n_components=300, random_state=42)
X_train = reduc.fit_transform(X_train)
elif args.reduction == 'CHI250':
reduc = SelectKBest(chi2, k=50)
X_train = reduc.fit_transform(X_train, y_train)
elif args.reduction == 'CHI2100':
reduc = SelectKBest(chi2, k=100)
X_train = reduc.fit_transform(X_train, y_train)
print(" Done!")
print(' New shape of training matrix: ', X_train.shape)
jobs = 2
paramGrid = []
nIter = 100
crossV = 5
print("Defining randomized grid search...")
if args.classifier == 'SVM':
# SVM
classifier = SVC()
if args.kernel == 'rbf':
paramGrid = {'C': scipy.stats.expon(scale=100),
# 'gamma': scipy.stats.expon(scale=.1),
'kernel': ['rbf'],
'class_weight': ['balanced', None]}
elif args.kernel == 'linear':
paramGrid = {'C': scipy.stats.expon(scale=100),
'kernel': ['linear'],
'class_weight': ['balanced', None]}
elif args.kernel == 'poly':
paramGrid = {'C': scipy.stats.expon(scale=100),
# 'gamma': scipy.stats.expon(scale=.1),
'degree': [2, 3],
'kernel': ['poly'],
'class_weight': ['balanced', None]}
myClassifier = model_selection.RandomizedSearchCV(classifier,
paramGrid, n_iter=nIter,
cv=crossV, n_jobs=jobs, verbose=3)
elif args.classifier == 'BernoulliNB':
# BernoulliNB
classifier = BernoulliNB()
paramGrid = {'alpha': scipy.stats.expon(scale=1.0)}
myClassifier = model_selection.RandomizedSearchCV(classifier, paramGrid, n_iter=nIter,
cv=crossV, n_jobs=jobs, verbose=3)
elif args.classifier == 'MultinomialNB':
# MultinomialNB
classifier = MultinomialNB()
paramGrid = {'alpha': scipy.stats.expon(scale=1.0)}
myClassifier = model_selection.RandomizedSearchCV(classifier, paramGrid, n_iter=nIter,
cv=crossV, n_jobs=jobs, verbose=3)
else:
print("Bad classifier")
exit()
print(" Done!")
print("Training...")
myClassifier.fit(X_train, y_train)
print(" Done!")
print("Testing (prediction in new data)...")
if args.reduction is not None:
X_test = reduc.transform(X_test)
y_pred = myClassifier.predict(X_test)
best_parameters = myClassifier.best_estimator_.get_params()
if args.classifier == "SVM":
confidence_scores = myClassifier.decision_function(X_test)
print(" Done!")
print("Saving report...")
with open(os.path.join(args.outputReportPath, args.outputReportFile), mode='w', encoding='utf8') as oFile:
oFile.write('********** EVALUATION REPORT **********\n')
oFile.write('Classifier: {}\n'.format(args.classifier))
oFile.write('Kernel: {}\n'.format(args.kernel))
oFile.write('Reduction: {}\n'.format(args.reduction))
oFile.write('Training best score : {}\n'.format(myClassifier.best_score_))
oFile.write('Accuracy: {}\n'.format(accuracy_score(y_test, y_pred)))
oFile.write('Precision: {}\n'.format(precision_score(y_test, y_pred, average='weighted')))
oFile.write('Recall: {}\n'.format(recall_score(y_test, y_pred, average='weighted')))
oFile.write('F-score: {}\n'.format(f1_score(y_test, y_pred, average='weighted')))
oFile.write('Confusion matrix: \n')
oFile.write(str(confusion_matrix(y_test, y_pred)) + '\n')
oFile.write('Classification report: \n')
oFile.write(classification_report(y_test, y_pred) + '\n')
oFile.write('Best parameters: \n')
for param in sorted(best_parameters.keys()):
oFile.write("\t%s: %r\n" % (param, best_parameters[param]))
if args.classifier == "SVM":
# if args.kernel == "linear":
# oFile.write('\nWeights assigned to the features: \n')
# oFile.write("{}\n".format(myClassifier.best_estimator_.coef_))
# oFile.write('Confidence scores: \n')
# oFile.write("{}\n".format(confidence_scores))
oFile.write('Number of support vectors per class: \n{}\n'.format(myClassifier.best_estimator_.n_support_))
oFile.write('Support vectors: \n{}\n'.format(myClassifier.best_estimator_.support_vectors_))
print(" Done!")
print("Training and testing done in: %fs" % (time() - t0))