countvect.py

cnt_vectorizer = CountVectorizer(dtype=np.float32,
            strip_accents='unicode', analyzer='word',token_pattern=r'\w{1,}',
            ngram_range=(1, 3),min_df=3)


# we fit count vectorizer to get ngrams from both train and test data.
cnt_vectorizer.fit(list(train_df.cleaned_text.values) + list(test_df.cleaned_text.values))

xtrain_cntv =  cnt_vectorizer.transform(train_df.cleaned_text.values) 
xtest_cntv = cnt_vectorizer.transform(test_df.cleaned_text.values)