# pass transformed data to nb for prediction
return self.style_nb_clf.predict(x_t)
if __name__ == "__main__":
from db.styles import Styles
from db.basewordcts import BaseWordFreq
from sklearn import cross_validation as c_v
from sklearn.metrics import classification_report, confusion_matrix
print 'Load baseline stop words'
baseline = BaseWordFreq()
baseline.load_all()
print 'Get reviews by style'
styles = Styles()
# get top n styles by review count
sty_ids = styles.review_counts(10).keys()
X = styles.beer_reviews_rollup(sty_ids, limit=0)
print 'Styles Retrieved: %s' % len(np.unique(X['style_id'].values))
print 'Beers Retrieved : %s' % len(np.unique(X['beer_id'].values))
print 'Total Rev Docs : %s' % len(X.index)
X_train, X_test, y_train, y_test = c_v.train_test_split(X['review'],
X['style_id'],
test_size=0.2,
random_state=0)
clf = StyleTfidfNB(max_features=None,
ngram_range=(1, 2),
def __init__(self): self.region = Locations() self.brewer = Brewers() self.style = Styles() self.beer = Beers()
