IMPLEMENTASI METODE MULTINOMIAL NAÏVE BAYES CLASSIFIER UNTUK ANALISIS SENTIMEN

Abstract

GO-JEK is an information technology-based service consisting of 3 main services, namely GO-JEK, GO-LIFE, and GO-PAY. As time goes by, customer complaints arise about the lack of satisfaction with GO-JEK services. One measure of GO-JEK customer service satisfaction can be acknowledged by way of analyzing sentiments on the data of public opinion twett submitted on the twitter official account, @gojekindonesia. In this research, sentiment analysis was carried out using the Naïve Bayes Classifier learning algorithm. The number of data twitts used is 9987. The twit data is labeled according to the class which includes positive, negative and neutral classes. Then the next process is data preprocessing consisting of cleansing, tokenization, filtering, stemming, and stopword removal. The evaluation method using 10-fold cross validation with the results obtained is precision value of 80%, recall of 80%, f1-score of 80%, maximum accuracy of 82% and average accuracy of 79%.