Analisis Sentimen Pencitraan Perguruan Tinggi di Yogyakarta Menggunakan Metode Naїve Bayes Classifier

Abstract

This research utilizes data from Twitter to analyze sentiment in Yogyakarta's universities using the Naïve Bayes Classifier method. The Naive Bayes Classifier method is one of the text classification methods based on the probability of keywords in comparing training and testing documents. The data used consists of tweets in Indonesian language with keywords from the top 10 universities in Yogyakarta based on webometrics, as well as four other relevant keywords about Yogyakarta that are frequently searched through Google. From the conducted research, there are 1710 data collected from Twitter, which are used for classification and categorized into 3 labels: positive, negative, and neutral. The data is divided into 70% for training and 30% for testing randomly. The result of sentiment analysis classification from the test data shows that 82.1% of the data is categorized as neutral, 14.8% as positive, and 3.1% as negative, with an accuracy value of 73%.