def createClassifier(ignoreTweets=False):
neg_ids = movie_reviews.fileids('neg')
pos_ids = movie_reviews.fileids('pos')
neg_sents = [(extractWords(movie_reviews.words(fileids=[f])), 'neg')
for f in neg_ids]
pos_sents = [(extractWords(movie_reviews.words(fileids=[f])), 'pos')
for f in pos_ids]
#if you dont want to process all tweets, just call :
(neg_pols, pos_pols) = getSentPolarities()
(neg_tweets, pos_tweets) = getTweets(ignoreTweets)
neg_sents = neg_sents + neg_tweets + neg_pols
pos_sents = pos_sents + pos_tweets + pos_pols
trainsizeneg = int(0.75 * len(neg_sents))
trainsizepos = int(0.75 * len(pos_sents))
all_train = neg_sents[:trainsizeneg] + pos_sents[:trainsizepos]
all_test = neg_sents[trainsizeneg:] + pos_sents[trainsizepos:]
# train size = 1500, test size = 500
s_analyzer = SentimentAnalyzer()
classifier = NaiveBayesClassifier.train(all_train)
print accuracy(classifier, all_test)
#classifier.show_most_informative_features()
return classifier
def classify(inputdir):
#filenames = os.listdir('d:\\shir\\')
filenames = os.listdir(inputdir)
feat_set = []
sets = []
for name in filenames:
# print name
lineno=0
path = os.path.join(inputdir, name)
sense = name.split('\\')[-1].split('.')[0]
print 'training', sense
file = codecs.open(path, 'r', 'utf-8')
allwords = []
for line in file:
if len(line.split())>2:
lineno+=1
line = line.strip()
words=[]
tags=[]
tokens = line.split()
for item in tokens:
if len(item.split('\\'))==2:
word=item.split('\\')[0]
tag= item.split('\\')[1]
words.append(word)
tags.append(tag)
allwords.append(word)
feat_set.append((bag_of_words(line),sense))
#feat_set.append((get_feature2(line),sense))
else:
words=[]
tags=[]
file.close()
random.shuffle(feat_set)
random.shuffle(feat_set)
#random.shuffle(feat_set)
train_data = train_feats(feat_set)
test_data = test_feats(feat_set)
#classifier= MaxentClassifier.train(train_data)
nb_classifier = NaiveBayesClassifier.train(train_data)
dt_classifier = DecisionTreeClassifier.train(train_data, entropy_cutoff=0.8, depth_cutoff=5, support_cutoff=30)
# pickle.dump(classifier, classifier_save_file)
entropy_classifier = MaxentClassifier.train(train_data,algorithm='iis', trace=0, max_iter=1, min_lldelta=0.5)
print "nb accuracy "+ str(accuracy(nb_classifier, test_data) * 100)
print "dt accuracy "+ str(accuracy(dt_classifier, test_data) * 100)
print "entropy accuracy "+ str(accuracy(entropy_classifier, test_data) * 100)
mv_classifier = MaxVoteClassifier(nb_classifier, dt_classifier, entropy_classifier)
print "max vote accuracy "+ str(accuracy(mv_classifier, test_data) * 100)
def train(cleanedDataCollection, tagPool): posSamples = [] negSamples = [] featuresets = [(extractFeatures(d,tagPool), c) for (d,c) in cleanedDataCollection] for sample in featuresets: if sample[1] == "trash": negSamples.append(sample) else: posSamples.append(sample) train_set = negSamples[10:]+posSamples[10:] test_set = negSamples[:10]+posSamples[:10] # classifier = nltk.NaiveBayesClassifier.train(train_set) # print(nltk.classify.accuracy(classifier, test_set)) # classifier.show_most_informative_features(5) # return classifier sk_classifier = SklearnClassifier(MultinomialNB()) sk_classifier.train(train_set) print "accuracy is: %s" % (accuracy(sk_classifier, test_set)) precision, recall, fMeasure = precision_recall_fmeasure(sk_classifier, test_set, "useful") print "precision is: %s" % (precision) print "recall is: %s" % (recall) print "F-measure is: %s" % (fMeasure) return sk_classifier
def main():
results = {'Topic': [], 'Precision': [], 'Recall': [], 'F-measure': []}
print('\nPreparing data...')
(train_set, test_set) = get_train_test_sets('data/content')
print('\nNB classifier training...')
classifier = NaiveBayesClassifier.train(train_set)
print('NB classifier is trained with {}% accuracy'.format(
round(accuracy(classifier, test_set) * 100, 1)))
refsets = collections.defaultdict(set)
testsets = collections.defaultdict(set)
for i, (feats, label) in enumerate(test_set):
refsets[label].add(i)
observed = classifier.classify(feats)
testsets[observed].add(i)
for topic in topics:
results['Topic'].append(topic)
results['Precision'].append(
round(precision(refsets[topic], testsets[topic]) * 100, 1))
results['Recall'].append(
round(recall(refsets[topic], testsets[topic]) * 100, 1))
results['F-measure'].append(
round(f_measure(refsets[topic], testsets[topic]) * 100, 1))
del classifier, train_set, test_set, refsets, testsets
gc.collect()
print(results)
def rte_classifier(trainer, features=rte_features):
"""
Classify RTEPairs
"""
train = [(pair, pair.value) for pair in nltk.corpus.rte.pairs(
['rte1_dev.xml', 'rte2_dev.xml', 'rte3_dev.xml'])]
test = [(pair, pair.value) for pair in nltk.corpus.rte.pairs(
['rte1_test.xml', 'rte2_test.xml', 'rte3_test.xml'])]
# Train up a classifier.
print('Training classifier...')
classifier = trainer([(features(pair), label) for (pair, label) in train])
# Run the classifier on the test data.
print('Testing classifier...')
acc = accuracy(classifier,
[(features(pair), label) for (pair, label) in test])
print('Accuracy: %6.4f' % acc)
test_label = [label for (pair, label) in test]
result = [classifier.classify(features(pair)) for (pair, label) in test]
print(metrics.accuracy_score(test_label, result))
print(metrics.classification_report(test_label, result))
# Return the classifier
return classifier
def train(self):
print 'Classifier Training in progress....'
poscutoff = len(self.positiveFeatures)
negcutoff = len(self.negativeFeatures)
print "Train Pos Cutoff: " + str(poscutoff) + " Train Neg Cutoff: " + str(negcutoff)
trainfeats = self.positiveFeatures[:poscutoff] + self.negativeFeatures[:negcutoff]
testfeats = self.test()
print 'Train on %d instances, test on %d instances' % (len(trainfeats), len(testfeats))
self.classifier = NaiveBayesClassifier.train(trainfeats)
print 'accuracy:', accuracy(self.classifier, testfeats)
refsets = collections.defaultdict(set)
testsets = collections.defaultdict(set)
for i, (feats, label) in enumerate(testfeats):
refsets[label].add(i)
observed = self.classifier.classify(feats)
#print label, observed
testsets[observed].add(i)
print 'pos precision:', nltk.metrics.precision(refsets['pos'], testsets['pos'])
print 'pos recall:', nltk.metrics.recall(refsets['pos'], testsets['pos'])
print 'pos F-measure:', nltk.metrics.f_measure(refsets['pos'], testsets['pos'])
print 'neg precision:', nltk.metrics.precision(refsets['neg'], testsets['neg'])
print 'neg recall:', nltk.metrics.recall(refsets['neg'], testsets['neg'])
print 'neg F-measure:', nltk.metrics.f_measure(refsets['neg'], testsets['neg'])
def model_test(classifier, test_features):
print('Model Accuracy: {0}'.format(accuracy(classifier, test_features)))
precisions, recalls, f_measure, conf_matrix = get_precision_recall_fmeasure_conf_matrix(classifier, test_features)
print('Precisions: {0}'.format(precisions))
print('Recalls: {0}'.format(recalls))
print('F-Measure: {0}'.format(f_measure))
print('Confusion Matrix: {0}'.format(conf_matrix))
def train_with_movie_db(self):
"""
Training possible with movie reviews
- this does not yield particularly good results
"""
self.use_movie_reviews = True
negids = movie_reviews.fileids('neg')
posids = movie_reviews.fileids('pos')
negfeats = [(self.feature_extraction_movie_reviews(movie_reviews.words(fileids=[f])),
"negative") for f in negids]
posfeats = [(self.feature_extraction_movie_reviews(movie_reviews.words(fileids=[f])),
"positive") for f in posids]
negcutoff = len(negfeats) * 3 / 4
poscutoff = len(posfeats) * 3 / 4
trainfeats = negfeats[:negcutoff] + posfeats[:poscutoff]
testfeats = negfeats[negcutoff:] + posfeats[poscutoff:]
DLOG("train on %d instances, test on %d instances" % (len(trainfeats), len(testfeats)))
self.classifier = NaiveBayesClassifier.train(trainfeats)
DLOG("accuracy: " + str(util.accuracy(self.classifier, testfeats)))
DLOG(self.classifier.show_most_informative_features())
def rte_classifier():
train_set = rte_corpus.pairs(
['rte1_dev.xml', 'rte2_dev.xml', 'rte3_dev.xml'])
test_set = rte_corpus.pairs(['rte1_test.xml'])
featurized_train_set = rte_featurize(train_set, True)
featurized_test_set = rte_featurize(test_set, False, test_id=0)
print('Training classifier...')
svm = SklearnClassifier(LinearSVC())
clf_svm = svm.train(featurized_train_set)
# clf_nb = nltk.NaiveBayesClassifier.train(featurized_train_set)
# clf_gis = MaxentClassifier.train(featurized_train_set, 'GIS')
# clf_iis = MaxentClassifier.train(featurized_train_set, 'IIS')
# clf_dt = SklearnClassifier(RandomForestClassifier(random_state=0)).train(featurized_train_set)
# clf_dt = DecisionTreeClassifier.train(featurized_train_set)
print('Testing classifier...')
# acc = m_accuracy(clf, featurized_test_set, test_set)
# acc_dt = accuracy(clf_dt, featurized_test_set)
# acc_gis = accuracy(clf_gis, featurized_test_set)
# acc_iis = accuracy(clf_iis, featurized_test_set)
acc_svm = accuracy(clf_svm, featurized_test_set)
# acc_nb = accuracy(clf_nb, featurized_test_set)
# print('rf Accuracy: %8.4f' % acc_dt)
print('svm Accuracy: %8.4f' % acc_svm)
# print('nb Accuracy: %8.4f' % acc_nb)
# print('gis Accuracy: %8.4f' % acc_gis)
# print('iis Accuracy: %8.4f' % acc_iis)
print '==================================='
def run_classifier_tests(classifier):
testfiles = [{
'F': 'C:/research/HIT_QIWEI/data/clean/label/F_r.txt'
}, {
'H': 'C:/research/HIT_QIWEI/data/clean/label/H_r.txt'
}, {
'O': 'C:/research/HIT_QIWEI/data/clean/label/O_r.txt'
}, {
'P': 'C:/research/HIT_QIWEI/data/clean/label/P_r.txt'
}, {
'S': 'C:/research/HIT_QIWEI/data/clean/label/S_r.txt'
}]
#testfiles = [{'performance': 'http://en.wikipedia.org/wiki/Performance_measurement'},
# {'resource': 'http://en.wikipedia.org/wiki/Resource_management'},
# {'risk': 'http://en.wikipedia.org/wiki/Risk_management'},
# {'strategic': 'http://en.wikipedia.org/wiki/Strategic_alignment'},
# {'value': 'http://en.wikipedia.org/wiki/Val_IT'},]
testfeats = []
for file in testfiles:
for sense, loc in file.iteritems():
for line in open(loc).read():
testfeats = testfeats + create_training_dict(line, sense)
acc = accuracy(classifier, testfeats) * 100
print 'accuracy: %.2f%%' % acc
def create_and_evaluate_classifier_10_fold(features):
"""
Uses 10 fold cross validation to create a classifier with naive bayes
for the given features. Evaluates the created classifier afterwards.
Results are printed to the console
:param features: List of features
"""
k_fold_validator = KFold(n_splits=10, shuffle=False)
accuracies = []
counter = 1
for train_index, test_index in k_fold_validator.split(features):
# print str(counter * 10) + '% cross-validation'
training_features = [features[i] for i in train_index]
test_features = [features[i] for i in test_index]
classifier = NaiveBayesClassifier.train(training_features)
accuracy = util.accuracy(classifier, test_features)
accuracies.append(accuracy)
counter = counter + 1
print 'Accuracy:', sum(accuracies) / float(len(accuracies)), '\n'
def rte_classifier(algorithm, sample_N=None):
from nltk.corpus import rte as rte_corpus
train_set = rte_corpus.pairs(
["rte1_dev.xml", "rte2_dev.xml", "rte3_dev.xml"])
test_set = rte_corpus.pairs(
["rte1_test.xml", "rte2_test.xml", "rte3_test.xml"])
if sample_N is not None:
train_set = train_set[:sample_N]
test_set = test_set[:sample_N]
featurized_train_set = rte_featurize(train_set)
featurized_test_set = rte_featurize(test_set)
# Train the classifier
print("Training classifier...")
if algorithm in ["megam"]: # MEGAM based algorithms.
clf = MaxentClassifier.train(featurized_train_set, algorithm)
elif algorithm in ["GIS", "IIS"]: # Use default GIS/IIS MaxEnt algorithm
clf = MaxentClassifier.train(featurized_train_set, algorithm)
else:
err_msg = str("RTEClassifier only supports these algorithms:\n "
"'megam', 'GIS', 'IIS'.\n")
raise Exception(err_msg)
print("Testing classifier...")
acc = accuracy(clf, featurized_test_set)
print("Accuracy: %6.4f" % acc)
return clf
def train(cleanedDataCollection, tagPool):
posSamples = []
negSamples = []
featuresets = [(extractFeatures(d, tagPool), c)
for (d, c) in cleanedDataCollection]
for sample in featuresets:
if sample[1] == "trash":
negSamples.append(sample)
else:
posSamples.append(sample)
train_set = negSamples[10:] + posSamples[10:]
test_set = negSamples[:10] + posSamples[:10]
# classifier = nltk.NaiveBayesClassifier.train(train_set)
# print(nltk.classify.accuracy(classifier, test_set))
# classifier.show_most_informative_features(5)
# return classifier
sk_classifier = SklearnClassifier(MultinomialNB())
sk_classifier.train(train_set)
print "accuracy is: %s" % (accuracy(sk_classifier, test_set))
precision, recall, fMeasure = precision_recall_fmeasure(
sk_classifier, test_set, "useful")
print "precision is: %s" % (precision)
print "recall is: %s" % (recall)
print "F-measure is: %s" % (fMeasure)
return sk_classifier
def rte_classifier(algorithm):
from nltk.corpus import rte as rte_corpus
train_set = rte_corpus.pairs(['rte1_dev.xml', 'rte2_dev.xml', 'rte3_dev.xml'])
test_set = rte_corpus.pairs(['rte1_test.xml', 'rte2_test.xml', 'rte3_test.xml'])
featurized_train_set = rte_featurize(train_set)
featurized_test_set = rte_featurize(test_set)
# Train the classifier
print('Training classifier...')
if algorithm in ['megam', 'BFGS']: # MEGAM based algorithms.
# Ensure that MEGAM is configured first.
check_megam_config()
clf = lambda x: MaxentClassifier.train(featurized_train_set, algorithm)
elif algorithm in ['GIS', 'IIS']: # Use default GIS/IIS MaxEnt algorithm
clf = MaxentClassifier.train(featurized_train_set, algorithm)
else:
err_msg = str(
"RTEClassifier only supports these algorithms:\n "
"'megam', 'BFGS', 'GIS', 'IIS'.\n"
)
raise Exception(err_msg)
print('Testing classifier...')
acc = accuracy(clf, featurized_test_set)
print('Accuracy: %6.4f' % acc)
return clf
def main():
'''
Main function of the boilerplate code is the entry point of the 'chitragoopt' executable script (defined in setup.py).
Use doctests, those are very helpful.
>>> main()
Hello
>>> 2 + 2
4
'''
lfeats = label_feats_from_corpus(movie_reviews)
train_feats, test_feats = split_label_feats(lfeats, split=0.75)
train_feats, test_feats = split_label_feats(lfeats, split=0.75)
# nb_classifier = NaiveBayesClassifier.train(train_feats)
print(sys.argv[1].split())
negfeat = bag_of_words(sys.argv[1].split())
f = open('my_classifier.pickle')
nb_classifier = pickle.load(f)
f.close()
print(accuracy(nb_classifier, test_feats))
print(nb_classifier.classify(negfeat))
for x in range(0, 50):
print(nb_classifier.classify(negfeat))
def rte_classifier(algorithm):
from nltk.corpus import rte as rte_corpus
train_set = rte_corpus.pairs(
['rte1_dev.xml', 'rte2_dev.xml', 'rte3_dev.xml'])
test_set = rte_corpus.pairs(
['rte1_test.xml', 'rte2_test.xml', 'rte3_test.xml'])
featurized_train_set = rte_featurize(train_set)
featurized_test_set = rte_featurize(test_set)
# Train the classifier
print('Training classifier...')
if algorithm in ['megam', 'BFGS']: # MEGAM based algorithms.
# Ensure that MEGAM is configured first.
check_megam_config()
clf = lambda x: MaxentClassifier.train(featurized_train_set, algorithm)
elif algorithm in ['GIS', 'IIS']: # Use default GIS/IIS MaxEnt algorithm
clf = MaxentClassifier.train(featurized_train_set, algorithm)
else:
err_msg = str("RTEClassifier only supports these algorithms:\n "
"'megam', 'BFGS', 'GIS', 'IIS'.\n")
raise Exception(err_msg)
print('Testing classifier...')
acc = accuracy(clf, featurized_test_set)
print('Accuracy: %6.4f' % acc)
return clf
def train(self):
"""Trains new sentimental analyzer model.
:return:
"""
data = self._get_train_data()
percentage_to_train = 0.9
neg_feats = [(word_feats(tokenize_clean_text(tweet[1])), self.NEGATIVE)
for tweet in data if tweet[0] == self.NEGATIVE]
pos_feats = [(word_feats(tokenize_clean_text(tweet[1])), self.POSITIVE)
for tweet in data if tweet[0] == self.POSITIVE]
del data
neg_cutoff = round(len(neg_feats) * percentage_to_train)
pos_cutoff = round(len(pos_feats) * percentage_to_train)
train_feats = neg_feats[:neg_cutoff] + pos_feats[:pos_cutoff]
test_feats = neg_feats[neg_cutoff:] + pos_feats[pos_cutoff:]
# Train Classifier.
print('train on %d instances, test on %d instances' %
(len(train_feats), len(test_feats)))
self._classifier = NaiveBayesClassifier.train(train_feats)
print('accuracy: ', accuracy(self._classifier, test_feats))
self._classifier.show_most_informative_features()
def test(self):
if not hasattr(self, 'test_sets'): self._split_features()
if not hasattr(self, 'classifiers'): self.train()
result = {}
for tag, classifier in self.classifiers.iteritems():
result[tag] = accuracy(classifier, self.test_sets[tag])
return result
def results(train, query_data, query_no_label, query_labels):
print '\nCalculating final results...'
megam_classifier = MaxentClassifier.train(train, 'megam') # build and train the maxent classifier
accu = accuracy(megam_classifier, query_data) # calculate the classification accuracy
predicted = megam_classifier.classify_many(query_no_label) # get a list of predicted labels
cm = confusion_matrix(query_labels, predicted) # build confusion matrix
return accu, cm
def implementMethods(self,sents,labelsData,clsent):
labelwords=[]
k=0
cl=self.featureList(sents,labelsData)
tr,te=self.setSplit(cl)
nb_classifier = NaiveBayesClassifier.train(tr)
print('Accuracy = '+str(accuracy(nb_classifier, te)*100)+'%')
return nb_classifier
def evaluate_accuracy(self):
'''Evaluate accuracy given a classifer model'''
accuracies = []
lfeats = self.label_feats_from_corpus()
for i in range(1, 10):
train_feats, test_feats, nb_classifier = self\
.__get_elements_for_classification(lfeats, train_number=i, classifying=False)
accuracies.append(accuracy(nb_classifier, test_feats))
return sum(accuracies)/len(accuracies)
def RunBayesNetwork(self, type_of_Feature_extractor):
#Bayes Network classifier, return accuracy
if type_of_Feature_extractor == 1:
#Format the positive and negative separately
formatted_pos_training = BNFormat.format_data(self.pos_training_data, "pos", BNFormat.Feature_extractor1)
formatted_neg_training = BNFormat.format_data(self.neg_training_data, "neg", BNFormat.Feature_extractor1)
#Same again but for the testing data
formatted_pos_testing = BNFormat.format_data(self.pos_testing_data, "pos", BNFormat.Feature_extractor1)
formatted_neg_testing = BNFormat.format_data(self.neg_testing_data, "neg", BNFormat.Feature_extractor1)
elif type_of_Feature_extractor == 2:
#Format the positive and negative separately
formatted_pos_training = BNFormat.format_data(self.pos_training_data, "pos", BNFormat.Feature_extractor2)
formatted_neg_training = BNFormat.format_data(self.neg_training_data, "neg", BNFormat.Feature_extractor2)
#Same again but for the testing data
formatted_pos_testing = BNFormat.format_data(self.pos_testing_data, "pos", BNFormat.Feature_extractor2)
formatted_neg_testing = BNFormat.format_data(self.neg_testing_data, "neg", BNFormat.Feature_extractor2)
elif type_of_Feature_extractor == 3:
#Format the positive and negative separately
formatted_pos_training = BNFormat.format_data(self.pos_training_data, "pos", BNFormat.Feature_extractor3)
formatted_neg_training = BNFormat.format_data(self.neg_training_data, "neg", BNFormat.Feature_extractor3)
#Same again but for the testing data
formatted_pos_testing = BNFormat.format_data(self.pos_testing_data, "pos", BNFormat.Feature_extractor3)
formatted_neg_testing = BNFormat.format_data(self.neg_testing_data, "neg", BNFormat.Feature_extractor3)
elif type_of_Feature_extractor == 4:
#Format the positive and negative separately
formatted_pos_training = BNFormat.format_data(self.pos_training_data, "pos", BNFormat.Feature_extractor4)
formatted_neg_training = BNFormat.format_data(self.neg_training_data, "neg", BNFormat.Feature_extractor4)
#Same again but for the testing data
formatted_pos_testing = BNFormat.format_data(self.pos_testing_data, "pos", BNFormat.Feature_extractor4)
formatted_neg_testing = BNFormat.format_data(self.neg_testing_data, "neg", BNFormat.Feature_extractor4)
else:
#Format the positive and negative separately
formatted_pos_training = BNFormat.format_data(self.pos_training_data, "pos")
formatted_neg_training = BNFormat.format_data(self.neg_training_data, "neg")
#Same again but for the testing data
formatted_pos_testing = BNFormat.format_data(self.pos_testing_data, "pos")
formatted_neg_testing = BNFormat.format_data(self.neg_testing_data, "neg")
#Combine them
formatted_training_data = formatted_pos_training + formatted_neg_training
#Combine them
formatted_testing_data = formatted_pos_testing + formatted_neg_testing
#Train on a list of reviews
nb_classifier = NaiveBayesClassifier.train(formatted_training_data)
#Print the features that the NB classifier found to be most important in making classifications
nb_classifier.show_most_informative_features()
#Test on another list of reviews
accuracy_ = accuracy(nb_classifier, formatted_testing_data)
return accuracy_
def evaluate_features(feature_extractor, N, only_acc=False):
from nltk.corpus import movie_reviews
from nltk.classify import NaiveBayesClassifier as naive
from nltk.classify.util import accuracy
from nltk.metrics import precision, recall, f_measure
from sys import stdout
negative = movie_reviews.fileids('neg')
positive = movie_reviews.fileids('pos')
negfeats = [(feature_extractor(movie_reviews.sents(fileids=[f])), 'neg')
for f in negative]
posfeats = [(feature_extractor(movie_reviews.sents(fileids=[f])), 'pos')
for f in positive]
negtrain, negtest = stratifiedSamples(negfeats, N)
postrain, postest = stratifiedSamples(posfeats, N)
trainfeats = negtrain + postrain
testfeats = negtest + postest
classifier = naive.train(trainfeats)
if only_acc: return accuracy(classifier, testfeats)
print 'accuracy: {}'.format(accuracy(classifier, testfeats))
# Precision, Recall, F-measure
from collections import defaultdict
refsets = defaultdict(set)
testsets = defaultdict(set)
for i, (feats, label) in enumerate(testfeats):
refsets[label].add(i)
observed = classifier.classify(feats)
testsets[observed].add(i)
print 'pos precision:', precision(refsets['pos'], testsets['pos'])
print 'pos recall:', recall(refsets['pos'], testsets['pos'])
print 'pos F-measure:', f_measure(refsets['pos'], testsets['pos'])
print 'neg precision:', precision(refsets['neg'], testsets['neg'])
print 'neg recall:', recall(refsets['neg'], testsets['neg'])
print 'neg F-measure:', f_measure(refsets['neg'], testsets['neg'])
stdout.flush()
classifier.show_most_informative_features()
return classifier
def evaluate_features(feature_extractor, N, only_acc=False):
from nltk.corpus import movie_reviews
from nltk.classify import NaiveBayesClassifier as naive
from nltk.classify.util import accuracy
from nltk.metrics import precision, recall, f_measure
from sys import stdout
negative = movie_reviews.fileids('neg')
positive = movie_reviews.fileids('pos')
negfeats = [(feature_extractor(movie_reviews.sents(fileids=[f])),
'neg') for f in negative]
posfeats = [(feature_extractor(movie_reviews.sents(fileids=[f])),
'pos') for f in positive]
negtrain, negtest = stratifiedSamples(negfeats, N)
postrain, postest = stratifiedSamples(posfeats, N)
trainfeats = negtrain + postrain
testfeats = negtest + postest
classifier = naive.train(trainfeats)
if only_acc: return accuracy(classifier, testfeats)
print 'accuracy: {}'.format(accuracy(classifier, testfeats))
# Precision, Recall, F-measure
from collections import defaultdict
refsets = defaultdict(set)
testsets = defaultdict(set)
for i, (feats, label) in enumerate(testfeats):
refsets[label].add(i)
observed = classifier.classify(feats)
testsets[observed].add(i)
print 'pos precision:', precision(refsets['pos'], testsets['pos'])
print 'pos recall:', recall(refsets['pos'], testsets['pos'])
print 'pos F-measure:', f_measure(refsets['pos'], testsets['pos'])
print 'neg precision:', precision(refsets['neg'], testsets['neg'])
print 'neg recall:', recall(refsets['neg'], testsets['neg'])
print 'neg F-measure:', f_measure(refsets['neg'], testsets['neg'])
stdout.flush()
classifier.show_most_informative_features()
return classifier
def dt_classify(filename):
raw_sample_stream = get_samples_stream(filename)
all_samples = list( binary_bow_feature(raw_sample_stream) )
# filter out two classes of outliers
# these two categories contain too few examples, so the word frequency in these two categories
# cannot reflect the true probability
# all_samples = [(features,aspect) for features,aspect in all_samples if aspect != common.AspectNothing and aspect != common.AspectBusiness]
test_sample_ratio = 0.25
train_samples,test_samples = split_samples(all_samples,test_sample_ratio)
print "training set has {} samples, test set has {} samples".format(len(train_samples),len(test_samples))
classifier = DecisionTreeClassifier.train(train_samples,binary=True, depth_cutoff=15,verbose=True)
print "training completes"
print "training accuracy: {}".format(accuracy(classifier,train_samples))
print "test accuracy: {}".format(accuracy(classifier,test_samples))
return classifier
def find_scores():
#Text formatting to classify
def format_text(text):
return ({word: True for word in nltk.word_tokenize(text)})
#Load positive categorized text
pos = []
with open("./pos.txt", encoding='ISO-8859-1') as f:
for i in f:
pos.append([
format_text(i.encode("utf-8").decode("unicode-escape")),
'positive'
])
#Load negative categorized text
neg = []
with open("./neg.txt", encoding='ISO-8859-1') as f:
for i in f:
neg.append([
format_text(i.encode("utf-8").decode("unicode-escape")),
'negative'
])
#Load negative categorized text
neu = []
with open("./neu.txt", encoding='ISO-8859-1') as f:
for i in f:
neu.append([
format_text(i.encode("utf-8").decode("unicode-escape")),
'neutre'
])
#Split data into training(80%) and testing(20%) sets
training_set = pos[:int((.80) * len(pos))] + neg[:int(
(.80) * len(neg))] + neu[:int((.80) * len(neu))]
test_set = pos[int((.80) * len(pos)):] + neg[int(
(.80) * len(neg)):] + neu[int((.80) * len(neu)):]
#Training classifier
classifier = NaiveBayesClassifier.train(training_set)
#Calculate scores
trueset = collections.defaultdict(set)
testset = collections.defaultdict(set)
#Test all test-set items using defined classifier
for i, (text, label) in enumerate(test_set):
trueset[label].add(i)
result = classifier.classify(text)
testset[result].add(i)
#accurays
return accuracy(classifier, test_set), f_measure(
trueset['positive'], testset['negative']), f_measure(
testset['negative'], trueset['positive']), f_measure(
testset['neutre'], trueset['positive']), f_measure(
testset['positive'], trueset['neutre']), f_measure(
testset['negative'],
trueset['neutre']), f_measure(testset['neutre'],
trueset['negative'])
def cross_validate():
training_set = load_training_set()
random.shuffle(training_set)
average = 0
cv = KFold(len(training_set), n_folds=10, indices=True, shuffle=False, random_state=None)
for traincv, evalcv in cv:
classifier = NaiveBayesClassifier.train(training_set[traincv[0]:traincv[len(traincv) - 1]])
acc = accuracy(classifier, training_set[evalcv[0]:evalcv[len(evalcv) - 1]])
print 'Range: ', evalcv[0], 'to', evalcv[len(evalcv) - 1]
print 'Accuracy: %4.2f' % acc
average += acc
print 'Average accuracy: %4.2f' % (average / 10)
def classifyReviews():
''' Perform sentiment classification on movie reviews '''
# Read the data from the file
data = pd.read_csv("data/movieReviews.csv")
# get the text of the positive and negative reviews only.
# positive and negative will be lists of strings
# For now we use only very positive and very negative reviews.
positive = getReviews(data, 4)
negative = getReviews(data, 0)
# Split each data set into training and testing sets.
# You have to write the function splitTrainTest
(posTrainText, posTestText) = splitTrainTest(positive, 0.8)
(negTrainText, negTestText) = splitTrainTest(negative, 0.8)
# Format the data to be passed to the classifier.
# You have to write the formatForClassifer function
posTrain = formatForClassifier(posTrainText, 'pos')
negTrain = formatForClassifier(negTrainText, 'neg')
# Create the training set by appending the pos and neg training examples
training = posTrain + negTrain
# Format the testing data for use with the classifier
posTest = formatForClassifier(posTestText, 'pos')
negTest = formatForClassifier(negTestText, 'neg')
# Create the test set
test = posTest + negTest
# Train a Naive Bayes Classifier
# Uncomment the next line once the code above is working
classifier = NaiveBayesClassifier.train(training)
# Uncomment the next two lines once everything above is working
print("Accuracy of the classifier is: " + str(accuracy(classifier, test)))
print("Accuracy of the positive is: " + str(accuracy(classifier, posTest)))
print("Accuracy of the negative is: " + str(accuracy(classifier, negTest)))
classifier.show_most_informative_features()
def naivebayes_classify(filename,filter_small_category):
raw_sample_stream = get_samples_stream(filename)
all_samples = list( binary_bow_feature(raw_sample_stream) )
if filter_small_category:
# filter out two classes of outliers
# these two categories contain too few examples, so the word frequency in these two categories
# cannot reflect the true probability
all_samples = [(features,aspect) for features,aspect in all_samples if aspect != common.AspectNothing and aspect != common.AspectBusiness]
test_sample_ratio = 0.25
train_samples,test_samples = split_samples(all_samples,test_sample_ratio)
print "training set has {} samples, test set has {} samples".format(len(train_samples),len(test_samples))
classifier = NaiveBayesClassifier.train(train_samples)
print "training completes"
print "########## training accuracy: {}".format(accuracy(classifier,train_samples))
print "########## test accuracy: {}".format(accuracy(classifier,test_samples))
classifier.show_most_informative_features(n=10)
return classifier
def run_classifier_tests(classifier):
testfiles = [{'traffic': 'traffic-training.txt'},
{'useless': 'useless-training.txt'}]
testfeats = []
for file in testfiles:
for sense, loc in file.iteritems():
for line in open(loc, 'r'):
testfeats = testfeats + create_training_dict(line, sense)
acc = accuracy(classifier, testfeats) * 100
print 'accuracy: %.2f%%' % acc
sys.exit()
def run_classifier_tests(classifier):
testfiles = [{'fruit': 'http://litfuel.net/plush/files/disambiguation/apple-fruit-training.txt'},
{'company': 'http://litfuel.net/plush/files/disambiguation/apple-company-training.txt'}]
testfeats = []
for file in testfiles:
for sense, loc in file.iteritems():
for line in urllib2.urlopen(loc):
testfeats = testfeats + create_training_dict(line, sense)
acc = accuracy(classifier, testfeats) * 100
print 'accuracy: %.2f%%' % acc
sys.exit()
def implementMethods(sents,labelsData,clsent):
labelwords=[]
k=0
for sent in sents:
labelwords.append((labelsData[k],nltk.tokenize.word_tokenize(sent)))
k=k+1
high_info_words=set(high_information_words(labelwords))
feat_det=lambda words:bag_of_words_in_set(words,high_info_words)
cl=featureList(sents,labelsData,feature_detector=feat_det)
tr,te=setSplit(cl)
nb_classifier = NaiveBayesClassifier.train(tr)
print('Accuracy = '+str(accuracy(nb_classifier, te)*100)+'%')
return nb_classifier
def run_classifier_tests(classifier):
testfiles = [{
'traffic': 'traffic-training.txt'
}, {
'useless': 'useless-training.txt'
}]
testfeats = []
for file in testfiles:
for sense, loc in file.iteritems():
for line in open(loc, 'r'):
testfeats = testfeats + create_training_dict(line, sense)
acc = accuracy(classifier, testfeats) * 100
print 'accuracy: %.2f%%' % acc
sys.exit()
def random_test(iterations, haiku_labeled, short_labeled, corpus_length):
import random
from nltk.classify import NaiveBayesClassifier #import directly from .py file while in appropriate directory (then use updated function, which returns features as list)
from nltk.classify.util import accuracy
random_scores = []
#need to mash the corpora together here, and then send the newly labeled corpora to the "for" loop
#switch labels for half of each corpora
haiku_a = haiku_labeled[0:int(len(haiku_labeled) / 2)]
haiku_b = haiku_labeled[int(len(haiku_labeled) / 2):]
for i in haiku_a:
i[1][1] = 'not-haiku'
poetry_a = short_labeled[0:int(len(short_labeled) / 2)]
poetry_b = short_labeled[int(len(short_labeled) / 2):]
for i in poetry_a:
i[1][1] = 'haiku'
#create new corpora based on these false labels
haiku_sample = haiku_b + poetry_a #i.e., real haiku and poetry labeled as haiku
poetry_sample = haiku_a + poetry_b
#now run the classification
for i in range(iterations):
#for i,k in zip(haiku_random[0::2], haiku_random[1::2]): #iterates through every other item
# i[1][1] = 'not-haiku'
#for i,k in zip(poetry_random[0::2], poetry_random[1::2]):
# i[1][1] = 'haiku'
haiku_random = []
poetry_random = []
haiku_random = random.sample(haiku_sample, corpus_length)
poetry_random = random.sample(poetry_sample, corpus_length)
cut_point = int(corpus_length / 4)
hfold1 = haiku_random[0:cut_point]
hfold2 = haiku_random[cut_point:(cut_point * 2)]
hfold3 = haiku_random[(cut_point * 2):(cut_point * 3)]
hfold4 = haiku_random[(cut_point * 3):]
pfold1 = poetry_random[0:cut_point]
pfold2 = poetry_random[cut_point:(cut_point * 2)]
pfold3 = poetry_random[(cut_point * 2):(cut_point * 3)]
pfold4 = poetry_random[(cut_point * 3):]
train_set = hfold1 + hfold2 + hfold3 + pfold1 + pfold2 + pfold3
test_set = hfold4 + pfold4
#train the classifier
nb_classifier = NaiveBayesClassifier.train([e[1] for e in train_set])
nb_classifier.labels()
#check accuracy of classifier and store accuracy measure
random_scores.append(
accuracy(nb_classifier, [el[1] for el in test_set]))
return random_scores
def trainDanger():
danger = []
with open("./anger.txt") as f:
for i in f:
danger.append([format_sentence(i), 'danger'])
calm = []
with open("./calm.txt") as f:
for i in f:
calm.append([format_sentence(i), 'calm'])
training = danger[:int((.8) * len(danger))] + calm[:int((.8) * len(calm))]
test = danger[int((.8) * len(danger)):] + calm[int((.8) * len(calm)):]
from nltk.classify import NaiveBayesClassifier
classifier = NaiveBayesClassifier.train(training)
from nltk.classify.util import accuracy
print("Test data accuracy" + str(accuracy(classifier, test)))
classifier.show_most_informative_features()
return classifier
def cross_validate(self):
"""
Performs cross validation by training the model on 90% of the
corpus then checking the accuracy on the remaining 10%.
"""
start = time.time()
feats = self.featureset()
offset = len(feats) / 10
random.shuffle(feats)
train = feats[:offset]
test = feats[offset:]
classifier, _ = self.train(train)
self.accuracy = accuracy(classifier, test)
self.validtime = time.time() - start
def rte_classifier(trainer, features=rte_features):
"""
Classify RTEPairs
"""
train = ((pair, pair.value) for pair in nltk.corpus.rte.pairs(['rte1_dev.xml', 'rte2_dev.xml', 'rte3_dev.xml']))
test = ((pair, pair.value) for pair in nltk.corpus.rte.pairs(['rte1_test.xml', 'rte2_test.xml', 'rte3_test.xml']))
# Train up a classifier.
print('Training classifier...')
classifier = trainer( [(features(pair), label) for (pair,label) in train] )
# Run the classifier on the test data.
print('Testing classifier...')
acc = accuracy(classifier, [(features(pair), label) for (pair,label) in test])
print('Accuracy: %6.4f' % acc)
# Return the classifier
return classifier
def cross_validate(self):
"""
Performs cross validation by training the model on 90% of the
corpus then checking the accuracy on the remaining 10%.
"""
start = time.time()
feats = self.featureset()
offset = len(feats) / 10
random.shuffle(feats)
train = feats[:offset]
test = feats[offset:]
classifier, _ = self.train(train)
self.accuracy = accuracy(classifier, test)
self.validtime = time.time() - start
def rte_classifier(trainer, features=rte_features):
"""
Classify RTEPairs
"""
train = ((pair, pair.value) for pair in nltk.corpus.rte.pairs(['rte1_dev.xml', 'rte2_dev.xml', 'rte3_dev.xml']))
test = ((pair, pair.value) for pair in nltk.corpus.rte.pairs(['rte1_test.xml', 'rte2_test.xml', 'rte3_test.xml']))
# Train up a classifier.
print('Training classifier...')
classifier = trainer( [(features(pair), label) for (pair,label) in train] )
# Run the classifier on the test data.
print('Testing classifier...')
acc = accuracy(classifier, [(features(pair), label) for (pair,label) in test])
print('Accuracy: %6.4f' % acc)
# Return the classifier
return classifier
def cross_validate():
training_set = load_training_set()
random.shuffle(training_set)
average = 0
cv = KFold(len(training_set),
n_folds=10,
indices=True,
shuffle=False,
random_state=None)
for traincv, evalcv in cv:
classifier = NaiveBayesClassifier.train(
training_set[traincv[0]:traincv[len(traincv) - 1]])
acc = accuracy(classifier,
training_set[evalcv[0]:evalcv[len(evalcv) - 1]])
print 'Range: ', evalcv[0], 'to', evalcv[len(evalcv) - 1]
print 'Accuracy: %4.2f' % acc
average += acc
print 'Average accuracy: %4.2f' % (average / 10)
def train(cleanedDataCollection, word_features, tagToBeTrained, high_info_wordSet): random.shuffle(cleanedDataCollection) featuresets = [(extractTfIdfFeatureOfADocument(d,word_features, high_info_wordSet), c) for (d,c) in cleanedDataCollection] train_set, test_set = featuresets[800:], featuresets[:800] # classifier = nltk.NaiveBayesClassifier.train(train_set) # print(nltk.classify.accuracy(classifier, test_set)) # classifier.show_most_informative_features(5) # return classifier sk_classifier = SklearnClassifier(MultinomialNB()) sk_classifier.train(train_set) print "accuracy is: %s" % (accuracy(sk_classifier, test_set)) precision, recall, fMeasure = precision_recall_fmeasure(sk_classifier, test_set, tagToBeTrained) print "precision is: %s" % (precision) print "recall is: %s" % (recall) print "F-measure is: %s" % (fMeasure) return sk_classifier
def trainAndTest(trFeatureSets, trLabels, teFeatureSets, teLabels, types, toPrint):
classifier = NaiveBayesClassifier.train(mapFeaturesToLabels(trFeatureSets, trLabels))
nuLabels = classifier.classify_many(teFeatureSets)
if toPrint:
print 'Test labels:', teLabels
print 'New labels:', nuLabels
print 'Accuracy: %.2f' % accuracy(classifier, mapFeaturesToLabels(teFeatureSets, teLabels))
# classifier.show_most_informative_features()
teIndices = getLabelIndices(teLabels, types)
nuIndices = getLabelIndices(nuLabels, types)
teCounts = Counter(teLabels)
nuCounts = Counter(nuLabels)
metrices = []
for type in types:
matches = len(teIndices[type] & nuIndices[type])
precision = 1.0 * matches / nuCounts[type] if nuCounts[type] > 0 else '-'
recall = 1.0 * matches / teCounts[type] if teCounts[type] > 0 else '-'
metrices += [[type, precision, recall]]
print tabulate(metrices, ['LABEL', 'PRECISION', 'RECALL'], tablefmt='fancy_grid', floatfmt='.2f')
def training():
pos_sen = open("positive.txt", 'r', encoding = 'latin-1').read()
neg_sen = open("negative.txt", 'r', encoding = 'latin-1').read()
emoji = open("emoji.txt",'r', encoding = 'latin-1').read()
pos_emoji = []
neg_emoji = []
for i in emoji.split('\n'):
exp = ''
if i[len(i)-2] == '-':
for j in range(len(i) - 2):
exp += i[j]
neg_emoji.append(( {exp : True}, 'negative'))
else:
for j in range(len(i)-1):
exp += i[j]
pos_emoji.append(( {exp : True}, 'positive'))
prev = [(features(words), 'positive') for words in pos_sen.split('\n')]
nrev = [(features(words), 'negative') for words in neg_sen.split('\n')]
pos_set = prev + pos_emoji
neg_set = nrev + neg_emoji
real_classifier = NaiveBayesClassifier.train(prev+nrev)
# SAVE IN FILE TO AVOID TRAIINING THE DATA AGAIN
save_doc = open("classifier.pickle", 'wb')
pickle.dump(real_classifier, save_doc)
save_doc.close()
#TO TEST ACCURACY OF CLASSIFIER UNCCOMMENT THE CODE BELOW
#ACCURACY : 78.1695423855964
ncutoff = int(len(nrev)*3/4)
pcutoff = int(len(prev)*3/4)
train_set = nrev[:ncutoff] + prev[:pcutoff] + pos_emoji + neg_emoji
test_set = nrev[ncutoff:] + prev[pcutoff:]
test_classifier = NaiveBayesClassifier.train(train_set)
print("Accuracy is : ", util.accuracy(test_classifier, test_set) * 100)
def main():
pid = movie_reviews.fileids('neg')
nid = movie_reviews.fileids('pos')
prev = [(features(movie_reviews.words(fileids=id)), 'positive')
for id in pid]
nrev = [(features(movie_reviews.words(fileids=id)), 'negative')
for id in nid]
ncutoff = int(len(nrev) * 3 / 4)
pcutoff = int(len(prev) * 3 / 4)
train_set = nrev[:ncutoff] + prev[:pcutoff]
test_set = nrev[ncutoff:] + prev[pcutoff:]
# NaiveBayesClassifier
classifier = NaiveBayesClassifier.train(train_set)
# Accuracy
print("Accuracy is : ", util.accuracy(classifier, test_set) * 100)
def makePrediction():
labels = movie_reviews.categories()
print("Labels for reviews are: {}\n".format(labels) )
labeled_words = [(label, movie_reviews.words(categories=[label])) for label in labels]
print("Labeled words:{}\n".format(labeled_words[:10]))
high_info_words = set(Toolbox.high_information_words(labeled_words))
print("High information words:{}\n".format(list(high_info_words)[:10]))
feat_det = lambda words: Toolbox.bag_of_words_in_set(words, high_info_words)
lfeats = Toolbox.label_feats_from_corpus(movie_reviews, feature_detector=feat_det)
train_feats, test_feats = Toolbox.split_label_feats(lfeats)
mv_classifier = ClassifierTrainer.trainClassifier(train_feats)
accuracyScore = accuracy(mv_classifier, test_feats)
print("Accuracy is {}".format(accuracyScore))
def train(self):
print 'Classifier Training in progress....'
poscutoff = len(self.positiveFeatures)
negcutoff = len(self.negativeFeatures)
print "Train Pos Cutoff: " + str(
poscutoff) + " Train Neg Cutoff: " + str(negcutoff)
trainfeats = self.positiveFeatures[:
poscutoff] + self.negativeFeatures[:
negcutoff]
testfeats = self.test()
print 'Train on %d instances, test on %d instances' % (len(trainfeats),
len(testfeats))
self.classifier = NaiveBayesClassifier.train(trainfeats)
print 'accuracy:', accuracy(self.classifier, testfeats)
refsets = collections.defaultdict(set)
testsets = collections.defaultdict(set)
for i, (feats, label) in enumerate(testfeats):
refsets[label].add(i)
observed = self.classifier.classify(feats)
#print label, observed
testsets[observed].add(i)
print 'pos precision:', nltk.metrics.precision(refsets['pos'],
testsets['pos'])
print 'pos recall:', nltk.metrics.recall(refsets['pos'],
testsets['pos'])
print 'pos F-measure:', nltk.metrics.f_measure(refsets['pos'],
testsets['pos'])
print 'neg precision:', nltk.metrics.precision(refsets['neg'],
testsets['neg'])
print 'neg recall:', nltk.metrics.recall(refsets['neg'],
testsets['neg'])
print 'neg F-measure:', nltk.metrics.f_measure(refsets['neg'],
testsets['neg'])
def trainLove():
#Delete common chars like comma
regex = re.compile('[\[,\.!?—\]]')
pos = []
with open("./pos_rom.txt") as f:
for i in f:
pos.append([format_sentence(regex.sub('', i.lower())), 'pos'])
neg = []
with open("./neg_rom.txt") as f:
for i in f:
neg.append([format_sentence(regex.sub('', i.lower())), 'neg'])
# next, split labeled data into the training and test data
training = pos[:int((.8) * len(pos))] + neg[:int((.8) * len(neg))]
test = pos[int((.8) * len(pos)):] + neg[int((.8) * len(neg)):]
from nltk.classify import NaiveBayesClassifier
classifier = NaiveBayesClassifier.train(training)
from nltk.classify.util import accuracy
print("Test data accuracy" + str(accuracy(classifier, test)))
classifier.show_most_informative_features()
data = []
#with open("./raw.txt", 'r', encoding='utf-8') as f:
# data=f.read()
return classifier
def main():
#print(movie_reviews)
##Importing The movie_reviews dataset pos and neg review
pid = movie_reviews.fileids('neg')
nid = movie_reviews.fileids('pos')
prev = [(features(movie_reviews.words(fileids = id)), 'positive') for id in pid]
nrev = [(features(movie_reviews.words(fileids = id)), 'negative') for id in nid]
ncutoff = int(len(nrev)*3/4)
pcutoff = int(len(prev)*3/4)
##Importing The movie_reviews dataset pos and neg review
#Training and Testing Sets
train_set = nrev[:ncutoff] + prev[:pcutoff]
test_set = nrev[ncutoff:] + prev[pcutoff:]
#Training and Testing Sets
# Training Text Classification Model and Evaluating The Model
classifier = NaiveBayesClassifier.train(train_set)
# Accuracy
print ("Accuracy is : ", util.accuracy(classifier, test_set) * 100)
def split_label_feats(lfeats, split=0.90):
train_feats = []
test_feats = []
for label, feats in lfeats.iteritems():
random.shuffle(feats, random.random)
cutoff = int(len(feats) * split)
train_feats.extend([(feat, label) for feat in feats[:cutoff]])
test_feats.extend([(feat, label) for feat in feats[cutoff:]])
return train_feats, test_feats
reader.categories()
lfeats = label_feats_from_corpus(reader)
lfeats.keys()
train_feats, test_feats = split_label_feats(lfeats)
len(train_feats)
len(test_feats)
from nltk.classify import NaiveBayesClassifier
nb_classifier = NaiveBayesClassifier.train(train_feats)
nb_classifier.labels()
from nltk.classify.util import accuracy
accuracy(nb_classifier, test_feats)
from nltk.classify.util import accuracy def bag_of_words(words): return dict([(word, True) for word in words]) train_feats =[]; test_feats=[]; #print(movie_reviews.categories()); #list categories #There are 1000 positive files and 1000 negative files for label in movie_reviews.categories(): #print [label]; #display neg,pos i=0; for fileid in movie_reviews.fileids([label]): i=i+1; words=movie_reviews.words(fileid); bag=bag_of_words(words); features=[bag,label]; if i < 750: train_feats.append(features); else: test_feats.append(features); #print len(train_feats); classifier = NaiveBayesClassifier.train(train_feats); #Naive bayes classifier #test the data posi=bag_of_words(['kate','winslet','is','accessible']); nega=bag_of_words(['the','plot','was','ludicrous']); print classifier.classify(nega); print classifier.classify(posi); #display accuracy print(accuracy(classifier,test_feats));
classifier = nltk.NaiveBayesClassifier.train(training_set)
def train(labeled_featuresets, estimator=ELEProbDist):
# Create the P(label) distribution
label_probdist = estimator(label_freqdist)
# Create the P(fval|label, fname) distribution
feature_probdist = {}
return NaiveBayesClassifier(label_probdist, feature_probdist)
#print "Most informative features"
#print classifier.show_most_informative_features(32)
print "-"*50
tweet = 'Multiple users on one machine never seemed to behave like youd expect'
print tweet
print "___/^"
#print extract_features(tweet.split())
print classifier.classify(extract_features(tweet.split()))
print "-"*50
print listare[0]
for texte in listare:
classifier.classify(extract_features(texte[0]))
acc = accuracy(classifier, training_set) * 100
print 'accuracy: %.2f%%' % acc
feature[word] = True
pdata.append((feature, 'POSITIVE'))
ndata = []
fileids = nc.movie_reviews.fileids('neg')
for fileid in fileids:
feature = {}
words = nc.movie_reviews.words(fileid)
for word in words:
feature[word] = True
ndata.append((feature, 'NEGATIVE'))
pnumb, nnumb = \
int(0.8 * len(pdata)), int(0.8 * len(ndata))
train_data = pdata[:pnumb] + ndata[:nnumb]
test_data = pdata[pnumb:] + ndata[nnumb:]
model = cf.NaiveBayesClassifier.train(train_data)
ac = cu.accuracy(model, test_data)
print('%.2f%%' % round(ac * 100, 2))
reviews = [
'It is an amazing movie.',
'This is a dull movie. I would never recommand it to anyone.',
'The cinemagraphy is pretty great in this movie',
'The direction was terrible and the story was all over the place'
]
sents, probs = [], []
for review in reviews:
feature = {}
words = review.split(' ')
for word in words:
feature[word] = True
pcls = model.prob_classify(feature)
sent = pcls.max()
print 'training a multi-binary %s classifier' % args.algorithm classifier = MultiBinaryClassifier.train(labels, train_feats, trainf, **train_kwargs) else: if args.trace: print 'training a %s classifier' % args.algorithm classifier = trainf(train_feats, **train_kwargs) ################ ## evaluation ## ################ if not args.no_eval: if not args.no_accuracy: print 'accuracy: %f' % accuracy(classifier, test_feats) if args.multi and args.binary and not args.no_masi_distance: print 'average masi distance: %f' % (scoring.avg_masi_distance(classifier, test_feats)) if not args.no_precision or not args.no_recall or not args.no_fmeasure: if args.multi and args.binary: refsets, testsets = scoring.multi_ref_test_sets(classifier, test_feats) else: refsets, testsets = scoring.ref_test_sets(classifier, test_feats) for label in labels: ref = refsets[label] test = testsets[label] if not args.no_precision:
def wordsInCorpus(corpus):
words = list()
for line in corpus:
for word in line.split():
if word not in stopWords:
words.append(normalize(word))
return words
posCorpus = open("finalPositiveCorpus.txt", "r")
poswords = wordsInCorpus(posCorpus)
negCorpus = open("finalNegativeCorpus.txt", "r")
negwords = wordsInCorpus(negCorpus)
#list of tuples (word, label) for every non-stop word that occurs in each corpus
labeled_features = ([(word, 'pos') for word in poswords] + [(word, 'neg') for word in negwords])
import random
random.shuffle(labeled_features)
cutOff = len(labeled_features) * 3/4
train_set, test_set = labeled_features[cutOff:], labeled_features[:cutOff]
#this is where the script crashes
#This WOULD work if NaiveBayesClassifier.train() worked on lists
#But it only works on dictionaries, contrary to what documentation says
classifier = NaiveBayesClassifier.train(train_set)
print 'accuracy:', accuracy(classifier, test_set)
arquivoClassificador.close()
arquivoClassificados = open('classificados.json')
classificados = ujson.load(arquivoClassificados)
arquivoClassificados.close()
sentimentos = {}
featuresClassificados = []
comeco = datetime.now()
for resposta in classificados:
texto = resposta['corpo']
frases = tokenizerFrases.tokenize(texto)
feature = {}
for frase in frases:
palavras = tokenizerPalavras.tokenize(frase)
palavras = [palavra for palavra in palavras if palavra not in stopWords]
for palavra in palavras:
feature[palavra] = True
sentimentos[texto] = (resposta, classificador.classify(feature))
featuresClassificados.append((feature, resposta['sentimento']))
tempo = datetime.now() - comeco
arquivoMedicoes = open('medicoes_analise_sequencial.txt', 'w')
arquivoMedicoes.write('Tempo de Execução = ' + str(tempo) + '\nPrecisão = {0:.2f}%'.format(accuracy(classificador, featuresClassificados) * 100))
arquivoMedicoes.close()
arquivoResultados = open('resultados_sem_stopwords.csv', 'w', newline='')
w = writer(arquivoResultados, delimiter=',')
linhas = [['Resposta', 'Pontos', 'Sentimento - Naive Bayes', 'Sentimento - AlchemyAPI']]
for texto in sentimentos.keys():
tupla = sentimentos[texto]
resposta = tupla[0]
linhas.append([texto, resposta['pontos'], tupla[1], resposta['sentimento']])
w.writerows(linhas)
arquivoResultados.close()
def word_features(words):
return dict([(word, True) for word in words])
# Get all the movie reviews with positive data set and negative data set
posRev = movie_reviews.fileids('pos')
negRev = movie_reviews.fileids('neg')
# Mark the words in data set as positive and negative:
posWords = [(word_features(movie_reviews.words(fileids=[f])), 'pos') for f in posRev]
negWords = [(word_features(movie_reviews.words(fileids=[f])), 'neg') for f in negRev]
# Set cut off for separating the training data and the testing data:
posCutoff = len(posWords) * 50 / 100
negCutoff = len(negWords) * 50 / 100
# Fill the training data and the testing data with positive and negative data set:
TestRev = posWords[posCutoff:] + negWords[negCutoff:]
Test_set = array(TestRev)
TrainRev = posWords[:posCutoff] + negWords[:negCutoff]
Train_set = array(TrainRev)
print 'train on %d instances, test on %d instances' % (len(Train_set), len(Test_set))
# Call Maximum Entropy classifier to classify the training data:
algo = MaxentClassifier.ALGORITHMS[0]
classifier = MaxentClassifier.train(Train_set, algorithm=algo, max_iter=3)
classifier.show_most_informative_features(10)
# Print the algorithm accuracy
print 'Accuracy is', util.accuracy(classifier, Test_set)
elif sentimento == 'negativo':
negativos.append(resposta)
else:
neutros.append(resposta)
threads = []
comeco = datetime.now()
for resposta in classificados:
thread = ThreadSentimento(resposta)
threads.append(thread)
thread.start()
for thread in threads:
thread.join()
tempo = datetime.now() - comeco
iteracao = iteracao + 1
arquivoMedicoes = open('medicoes_analise_threads_' + str(iteracao) + '.txt', 'w')
precisao = accuracy(classificador, featuresClassificados) * 100
arquivoMedicoes.write('Tempo de Execução = ' + str(tempo) + '\nPrecisão = {0:.2f}%'.format(precisao))
arquivoMedicoes.close()
features = resultadoPositivos.get() + resultadoNegativos.get() + resultadosNeutros.get()
pool1.terminate()
pool1.close()
pool2.terminate()
pool2.close()
pool3.terminate()
pool3.close()
if precisao > 50:
features.extend(featuresClassificados)
shuffle(features)
classificador = NaiveBayesClassifier.train(features)
arquivoClassificador = open('classificador.pickle', 'wb')
dump(classificador, arquivoClassificador, protocol=HIGHEST_PROTOCOL)
