-
Notifications
You must be signed in to change notification settings - Fork 0
/
classificationBenchmarker.py
285 lines (236 loc) · 11.3 KB
/
classificationBenchmarker.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
import datasetRetriever as dr
from json2html import *
import json
import nltk
import numpy as np
import sklearn
import sys
import time
from functools import partial
from nltk.metrics.scores import f_measure
from sklearn.cluster import KMeans
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics import accuracy_score, f1_score
from sklearn.naive_bayes import GaussianNB
from sklearn.neighbors import KNeighborsClassifier
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier as DTC
from textblob.classifiers import NaiveBayesClassifier, DecisionTreeClassifier
from tweetTemplate import *
def main():
t0 = time.time()
# DRY
dr.init(200)
results = []
rmax = 1 # Intended for k-fold cross validation. Vary as k
for r in range(0, rmax):
sys.stdout.write("\rOverall Progress: %d%%" % int(r * 100 / rmax))
sys.stdout.flush()
print
# Get features, labelled bigrams, test and training sets
dr.prepareDatasets()
bigramFeatures, labelledBigrams = dr.getTrainingDataset()
featureExtractor = partial(extractFeatures, gramFeatures = bigramFeatures)
train_features, train_labels = dr.getTrainingSet()
test_features, test_labels = dr.getTestSet()
training_set = nltk.classify.apply_features(featureExtractor, labelledBigrams)
labelledTest = dr.getTestDataset()
test_set = nltk.classify.apply_features(featureExtractor, labelledTest)
print "training_set: {0}, test_set: {1}".format(str(len(training_set)), str(len(test_set)))
# Best results are obtained for NB (manual IDF) and DTrees
for i in range(1,4):
res = classifyKNN_sklearn((train_features, train_labels), (test_features, test_labels), i)
results.append(res)
res = classifyNB_sklearn((train_features, train_labels), (test_features, test_labels))
results.append(res)
results.append(mClassifyNB(training_set, test_set))
results.append(mClassifyNB_sklearn(training_set, test_set))
for i in range(1, 4):
results.append(mClassifyKNN(training_set, test_set, i))
results.append(classifyDT_sklearn((train_features, train_labels), (test_features, test_labels), 8))
#results.append(classifyKMC_sklearn((train_features, train_labels), (test_features, test_labels)))
results.append(classifySVM_sklearn((train_features, train_labels), (test_features, test_labels)))
resultsFile = open("results.json", 'w')
json.dump(list(results), resultsFile)
resultsFile.close()
ld_train = dr.getLabelledDataset("../data/training_set.tds")
ld_test = dr.getLabelledDataset("../data/test_set.tds")
# Text blob is know to outperform current implementation, leave it out for now
#classifyNB_tb(ld_train, ld_test)
#classifyDT_tb(ld_train, ld_test)
htmlTemplateBegin = "<!DOCTYPE html><html><head><title>Sarcastic results</title></head><body><h2>Sarcasm Detection Results</h2>"
htmlTemplateEnd = "</body></html>"
resHtml = open("Benchmark.html", 'w')
resHtml.write(htmlTemplateBegin)
resHtml.write(json2html.convert(json={'results': results}))
resHtml.write(htmlTemplateEnd)
resHtml.close()
print "Execution complete."
print "Total time of execution: {0} mins.".format(str(int((time.time() - t0) / 60)))
return
# Naive Bayes Classifier - nltk implementation - using manual IDF
def mClassifyNB(training_set, test_set):
print "Naive Bayes Classifier"
# print "Training..."
classifier = nltk.NaiveBayesClassifier.train(training_set)
# print "Training complete."
tr_acc = nltk.classify.accuracy(classifier, training_set)
te_acc = nltk.classify.accuracy(classifier, test_set)
results = classifier.classify_many([fs for (fs, lbl) in test_set])
reference = [lbl for (fs, lbl) in test_set]
f1_score = f_measure(set(reference), set(results))
return generateJson("Naive Bayes - manual IDF", tr_acc, te_acc, f1_score)
# k Nearest Neighbors - sklearn implementation
def classifyKNN_sklearn(training_set, test_set, n):
print "K-Nearest Neighbor Classifier with {0} nearest neighbors.".format(str(n))
train_features, train_labels = training_set
test_features, test_labels = test_set
classifier = KNeighborsClassifier(n_neighbors=n)
# print "Training..."
classifier.fit(train_features, train_labels)
# print "Training complete."
tr_acc = accuracy_score(classifier.predict(train_features), train_labels)
te_acc = accuracy_score(classifier.predict(test_features.todense()), test_labels)
f_score = f1_score(test_labels, classifier.predict(test_features.todense()), average='micro')
return generateJson("k-NN, k = %d" % n, tr_acc, te_acc, f_score)
# Naive Bayes classifier - sklearn implementation
def classifyNB_sklearn(training_set, test_set):
print "Naive Bayes (sklearn) Classifier"
classifier = GaussianNB()
# print "Training."
train_features, train_labels = training_set
test_features, test_labels = test_set
classifier.fit(train_features, train_labels)
# print "Training complete."
tr_acc = accuracy_score(classifier.predict(train_features), train_labels)
te_acc = accuracy_score(classifier.predict(test_features.todense()), test_labels)
f_score = f1_score(test_labels, classifier.predict(test_features.todense()), average='micro')
return generateJson("Naive Bayes (sklearn)", tr_acc, te_acc, f_score)
def classifySVM_sklearn(training_set, test_set):
print "SVM (sklearn) Classifier"
classifier = SVC(C=1000.0)
# print "Training."
train_features, train_labels = training_set
test_features, test_labels = test_set
classifier.fit(train_features, train_labels)
# print "Training complete."
tr_acc = accuracy_score(classifier.predict(train_features), train_labels)
te_acc = accuracy_score(classifier.predict(test_features.todense()), test_labels)
f_score = f1_score(test_labels, classifier.predict(test_features.todense()), average='micro')
return generateJson("SVM (sklearn)", tr_acc, te_acc, f_score)
# Decision Trees - sklearn implementation
def classifyDT_sklearn(training_set, test_set, i):
print "Decision trees (sklearn) Classifier ({0})".format(str(i))
print "Training."
# Decision Trees: min-split-sample = 8 and max-features = 200
classifier = DTC(min_samples_split=3)
train_features, train_labels = training_set
test_features, test_labels = test_set
classifier.fit(train_features, train_labels)
#print "Training complete."
tr_acc = accuracy_score(classifier.predict(train_features), train_labels)
te_acc = accuracy_score(classifier.predict(test_features), test_labels)
f_score = f1_score(test_labels, classifier.predict(test_features.todense()), average='micro')
return generateJson("Decision Trees (sklearn)", tr_acc, te_acc, f_score)
# K Means Clustering classifier - sklearn implementation
def classifyKMC_sklearn(training_set, test_set):
print "K Means Clustering (sklearn) Classifier"
classifier = KMeans(n_clusters=5)
train_features, train_labels = training_set
test_features, test_labels = test_set
# print "Training."
classifier.fit(train_features)
# print "Training complete."
tr_acc = calcAcc(train_features, train_labels, classifier)
te_acc = calcAcc(test_features, test_labels, classifier)
f_score = f1_score(test_labels, classifier.predict(test_features.todense()), average='micro')
return generateJson("K Means Clustering (sklearn)", tr_acc, te_acc, f_score)
# Naive Bayes Classifier - TextBlob implementation
def classifyNB_tb(training_set, test_set):
print "TextBlob Naive Bayes."
cl = NaiveBayesClassifier(training_set)
print "Trained"
print "Train set accuracy: " + str(cl.accuracy(training_set))
print "Test set accuracy: " + str(cl.accuracy(test_set))
# Decision trees - TextBlob implementation
def classifyDT_tb(training_set, test_set):
print "TextBlob Decision Trees."
cl = DecisionTreeClassifier(training_set)
print "Trained"
print "Train set accuracy: " + str(cl.accuracy(training_set))
print "Test set accuracy: " + str(cl.accuracy(test_set))
# Naive Bayes Classifier (sklearn implementation) - using manual IDF
def mClassifyNB_sklearn(training_set, test_set):
print "Naive Bayes (sklearn) Classifier (m)"
classifier = GaussianNB()
train_features, train_labels, test_features, test_labels = \
getFeaturesLabels(training_set, test_set)
# print "Training."
classifier.fit(train_features, train_labels)
# print "Training complete."
tr_acc = accuracy_score(classifier.predict(train_features), train_labels)
te_acc = accuracy_score(classifier.predict(test_features), test_labels)
f_score = f1_score(test_labels, classifier.predict(test_features), average='micro')
return generateJson("Naive Bayes (sklearn) - manual IDF", tr_acc, te_acc, f_score)
# k Nearest Neighbors - using manual IDF
def mClassifyKNN(training_set, test_set, n):
print "K-Nearest Neighbor Classifier with {0} nearest neighbors. (m)".format(str(n))
train_features, train_labels, test_features, test_labels = \
getFeaturesLabels(training_set, test_set)
classifier = KNeighborsClassifier(n_neighbors=n)
# print "Training..."
classifier.fit(train_features, train_labels)
# print "Training complete."
tr_acc = accuracy_score(classifier.predict(train_features), train_labels)
te_acc = accuracy_score(classifier.predict(test_features), test_labels)
f_score = f1_score(test_labels, classifier.predict(test_features), average='micro')
return generateJson("k-NN, k = %d - manual IDF" % n, tr_acc, te_acc, f_score)
# Extract features from a bigram
def extractFeatures(doc, gramFeatures):
document = set(doc)
features = {}
for gram in gramFeatures:
features['contains(({0}, {1}))'.format(gram[0], gram[1])] = (gram in document)
return features
# Extract features and labels from training ans test sets, given a labellled set
def getFeaturesLabels(training_set, test_set):
train_features = []
train_labels = []
for features, label in training_set:
tf = []
for feature in features:
if features[feature]:
tf.append(1.)
else: tf.append(0.)
train_features.append(tf)
train_labels.append(label)
test_features = []
test_labels = []
for feature, label in test_set:
tf = []
for feature in features:
if features[feature]:
tf.append(1.)
else: tf.append(0.)
test_features.append(tf)
test_labels.append(label)
return train_features, train_labels, test_features, test_labels
# K Means classifier returns an array as a prediction, and doesn't work properly
# with accuracy_score. Hence, a util
def calcAcc(features, labels, classifier):
j = 0.
for i in range(len(labels)):
p = str(classifier.predict(features[i])[0])
l = labels[i]
if p == l:
j += 1.
return float(j / len(labels))
def generateJson(algorithm, train_acc, test_acc, f_score):
return {
'Algorithm': algorithm,
'Training Set accuracy' : "%.2f%%" % (train_acc * 100),
'Test Set accuracy' : "%.2f%%" % (test_acc * 100),
'F - Score' : "%.2f%%" % (f_score * 100)
}
if __name__ == "__main__":
main()