def fit(self, cleaned_data, terms, target, stopwords, idf, weight=None):
self.cleaned_data = cleaned_data
self.terms = terms
self.target = target
if weight == None:
weighting = Weighting(self.cleaned_data, self.terms)
self.weighted_terms = weighting.get_tf_idf_weighting()
else:
self.weighted_terms = weight
self.idf = idf
self.stopwords = stopwords
for i in range(len(self.cleaned_data)):
total_word = 0
for term in self.terms:
total_word += self.weighted_terms[term][i]
self.total.append(total_word)
for term in self.terms:
self.con_prob_negative.append(
self.calculate_probability_multinomial(term, 'Negatif'))
self.con_prob_neutral.append(
self.calculate_probability_multinomial(term, 'Netral'))
self.con_prob_positive.append(
self.calculate_probability_multinomial(term, 'Positif'))
self.likelihood = {}
indexKomentar = 0
for term in self.terms:
temp = []
temp.append(self.con_prob_negative[indexKomentar])
temp.append(self.con_prob_neutral[indexKomentar])
temp.append(self.con_prob_positive[indexKomentar])
self.likelihood[term] = temp
# buatprint = []
# buatprint.append(term)
# for t in temp:
# formatdulu = "{:.6f}".format(t)
# buatprint.append(str(formatdulu))
# print(";".join(buatprint))
indexKomentar += 1
self.prior_negative = self.getTotalDocumentWithSpecificCategory(
'Negatif') / self.getTotalDocument()
self.prior_neutral = self.getTotalDocumentWithSpecificCategory(
'Netral') / self.getTotalDocument()
self.prior_positive = self.getTotalDocumentWithSpecificCategory(
'Positif') / self.getTotalDocument()
result_data_file.close()
result_data_file = open(nt_filename, 'rb')
new_terms = pickle.load(result_data_file)
result_data_file.close()
result_data_file = open(rw_filename, 'rb')
removed_words = pickle.load(result_data_file)
result_data_file.close()
result_data_file = open(stopwords_filename, 'rb')
stopwords = pickle.load(result_data_file)
result_data_file.close()
weight = Weighting(new_cleaned_data, new_terms)
tfidf = weight.get_tf_idf_weighting()
idf = weight.get_idf()
nb = NBMultinomial()
nb.fit(new_cleaned_data, new_terms, data_train[i]["target"], stopwords, idf,
tfidf)
def predict(input_tweet):
prediction, negatif, netral, positif, used_terms = nb.predict(input_tweet)
result_dict = {}
result_dict["success"] = True
result_dict["type"] = "TBRS"
result_dict["data"] = {
"prediction": prediction,
"negatif": negatif,