def fit(self, table):
stemmed_text = stem(table.text)
self.tfidf = TfidfVectorizer(stop_words='english', max_features=self.max_features)
text_features = self.tfidf.fit_transform(stemmed_text)
text_features = text_features.toarray()
self.pca = PCA(150).fit(text_features)
self.avg_user = CategoryAverage()
self.avg_user = self.avg_user.fit(table.user_id, log(table.votes_useful + 1))
# scale for votes
votes = log(table.votes_useful + 1)
if self.use_scale:
self.scale = 1 / max(votes)
else:
self.scale = 1
return self
