Sentiment Analysis of Local Water Company Customer Using Naive Bayes Algorithm

The rapid evolution of technology has precipitated several transformations in the understanding of human life, including the widespread availability of diverse knowledge and solutions to previously acknowledged challenges. Researchers now harness the potential of X (Twitter), a social media platform, for conducting sentiment analysis studies aimed at elucidating feelings, attitudes, and public opinions. The employed processing technique takes the form of text mining. Furthermore, investigations were carried out utilizing the Naive Bayes Classifier algorithm, specifically focusing on Multinomial Naive Bayes. The experiment entailed utilizing imbalanced data and applying the Random Over Sampling method to both training and test data sets, with ratios ranging from 90:10 to 60:40. The assessment of the Naive Bayes algorithm revealed that the highest accuracy, reaching 76%, was achieved with a data distribution ratio of 80:20 for training and test data, coupled with random over sampling.