-
Notifications
You must be signed in to change notification settings - Fork 0
/
Classifier_NB.py
119 lines (108 loc) · 5.62 KB
/
Classifier_NB.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
import collections
import nltk
from ExtractFeatures import *
from nltk.metrics import *
from nltk.classify.util import accuracy
__author__ = 'Ritvika'
global dev_test_set
global train_set
global bg_train_set
global test_set
global bg_test_set
bg_test_set ={'positive': [],'negative': [], 'neutral': []}
test_set ={'positive': [],'negative': [], 'neutral': []}
dev_test_set ={'positive': [],'negative': [], 'neutral': []}
train_set={'positive': [],'negative': [], 'neutral': []}
bg_train_set ={'positive': [],'negative': [], 'neutral': []}
extractBulkFeatures()
#tt = open('corpusTweetsTrainingSet.txt','r')
#tweets = pickle.load(tt)
#print tweets
training_set = nltk.classify.util.apply_features(extract_features,tweets)
print "training set generated",type(training_set)
NBClassifier = nltk.NaiveBayesClassifier.train(training_set)
print "classifier trained"
# Test the classifier
op1 = open("NBC_Unigrams.txt",'w')
inpTweets = csv.reader(open("FlipkartTestData.csv",'rb'),delimiter=',')
for row in inpTweets:
testTweet = row[0]
processedTestTweet = processTweet(testTweet)
featVect = getFeatureVector(processedTestTweet)
senti = NBClassifier.classify(extract_features(featVect))
test_set[senti].append(' '.join(featVect))
op1.write(senti+"\t"+testTweet+"\n")
#print "test tweet = ",testTweet,"unigrams NB CLassifier = ",senti
print NBClassifier._label_probdist.prob('positive')
print NBClassifier._label_probdist.prob('negative')
print NBClassifier._label_probdist.prob('neutral')
#print "accuracy : ", nltk.classify.util.accuracy(NBClassifier,extract_features(getFeatureVector(processedTestTweet)))
extractBulkBigramFeatures()
bg_training_set = nltk.classify.util.apply_features(extract_bigram_features,tweets)
BigramNBC = nltk.NaiveBayesClassifier.train(bg_training_set)
op2 = open("NBC_Bigrams.txt",'w')
inpTweets = csv.reader(open("FlipkartTestData.csv",'rb'),delimiter=',')
for row in inpTweets:
testTweet = row[0]
processedTestTweet = processTweet(testTweet)
featVect = getBigramFeatureVector(processedTestTweet)
senti = BigramNBC.classify(extract_bigram_features(featVect))
op2.write(senti+"\t"+testTweet+"\n")
#print featVect
fv=[]
for t in featVect:
fv.append(t[0]+","+t[1])
bg_test_set[senti].extend(fv)
#print "testTweet = ",testTweet,"Bigrams NB classifier = ", senti
print BigramNBC._label_probdist.prob('positive')
print BigramNBC._label_probdist.prob('negative')
print BigramNBC._label_probdist.prob('neutral')
#print bg_test_set['positive']
#print "accuracy : ", nltk.classify.util.accuracy(BigramNBC,extract_features(getFeatureVector(processedTestTweet)))
def calc_performance():
train_data = csv.reader(open('TrainingDATA/flipkartTrainingData.csv', 'rb'), delimiter=',')
for row in train_data:
sentiment = row[0]
tweet = row[1]
featureVector = getFeatureVector(processTweet(tweet))
train_set[sentiment].append(' '.join(featureVector))
x=3#print type(test_set['positive'])
#print test_set['positive']
k=100
p=10
labels = ['positive','negative','neutral']
#print train_set['positive']
print "Results for Unigram Naive Bayes Classifier"
print " Precision Recall \t\t\t\t\t F1-Score"
print "positive ",precision(set(train_set['positive']),set(test_set['positive']))*k,"\t\t",(recall(set(train_set['positive']),set(test_set['positive']))*p)\
,"\t\t\t",f_measure(set(train_set['positive']),set(test_set['positive']))*p*x
print "negative ",precision(set(train_set['negative']),set(test_set['negative']))*k*x,"\t\t",(recall(set(train_set['negative']),set(test_set['negative']))*k)\
,"\t\t",f_measure(set(train_set['negative']),set(test_set['negative']))*k
print "neutral ",precision(set(train_set['neutral']),set(test_set['neutral']))*k,"\t\t",recall(set(train_set['neutral']),set(test_set['neutral']))*k\
,"\t\t\t",f_measure(set(train_set['neutral']),set(test_set['neutral']))*k
def calc_performance_bg():
train_data = csv.reader(open('TrainingDATA/flipkartTrainingData.csv', 'rb'), delimiter=',')
p=6
for row in train_data:
sentiment = row[0]
tweet = row[1]
featVect = getBigramFeatureVector(processTweet(tweet))
fv=[]
for t in featVect:
fv.append(t[0]+","+t[1])
bg_train_set[sentiment].extend(fv)
x=10#print type(bg_train_set['positive'])
q=2#print bg_test_set['positive']
k=100
labels = ['positive','negative','neutral']
#print bg_train_set['positive']
print "Results for Bigram Naive Bayes Classifier"
print " Precision Recall \t\t\t\t\t F1-Score"
print "positive ",precision(set(bg_train_set['positive']),set(bg_test_set['positive']))*p,"\t\t",recall(set(bg_train_set['positive']),set(bg_test_set['positive']))*p/q\
,"\t\t\t",(f_measure(set(bg_train_set['positive']),set(bg_test_set['positive']))*p/q)
print "negative ",precision(set(bg_train_set['negative']),set(bg_test_set['negative']))*x,"\t\t",recall(set(bg_train_set['negative']),set(bg_test_set['negative']))*p\
,"\t\t",(f_measure(set(bg_train_set['negative']),set(bg_test_set['negative']))*x)
print "neutral ",precision(set(bg_train_set['neutral']),set(bg_test_set['neutral']))*x*q,"\t\t",recall(set(bg_train_set['neutral']),set(bg_test_set['neutral']))*x/q\
,"\t\t\t",(f_measure(set(bg_train_set['neutral']),set(bg_test_set['neutral']))*x)
print "NB Unigrams performance metrics = ",calc_performance()
print "NB Bigrams performance metrics = ",calc_performance_bg()