def test_features(self): """tests bag of words features It appears to work well """ docs = self.source.find_clean(batch_size=1000) for ind, doc in enumerate(docs): print(cls.feature(doc)) if ind == 1: break
def test_evaluate(self):
ukr = self.source.find_clean({"subreddit" : "UkrainianConflict"}, limit=2500, batch_size=1000)
askr = self.source.find_clean({"subreddit" : "AskReddit"}, limit=2500, batch_size=1000)
alll = self.source.find_clean(limit=10000)
featuresets = [(cls.feature(doc), "YES") for doc in ukr]
featuresets.extend([(cls.feature(doc), "NO") for doc in askr])
random.shuffle(featuresets)
trainset, testset = featuresets[1250:], featuresets[:1250]
classifier = NaiveBayesClassifier.train(trainset)
f = open("./UkrainianConflictNVM","w")
for doc in alll:
del (doc["_id"])
truthiness = False
truthiness = classifier.classify(cls.feature(doc))
if truthiness:
f.write(json.dumps(doc) + "\n")
f.close()
print(nltk.classify.accuracy(classifier, testset))
def test_classifier(self):
positive = self.source.find_clean({"subreddit" : "UkrainianConflict"}, limit=2500, batch_size=1000)
other = self.source.find_clean(limit=2500,batch_size=1000)
classifier = cls.positive_naive_bayes(positive,other)
news = self.source.find_clean({"subreddit" : "news"}, limit=10)
f = open("./UkrainianConflict", "w")
for doc in news:
del (doc["_id"])
truthiness = False
truthiness = classifier.classify(cls.feature(doc))
if truthiness:
f.write(json.dumps(doc) + "\n")
f.close()
classifier.show_most_informative_features()