A Comparative Study of Machine Learning Models for Sentiment Analysis on User Reviews of the Desa Wisata Nusantara Application in the Google Play Store

In the swiftly changing digital environment, user reviews on platforms like the Google Play Store have emerged as a crucial factor influencing an application's success. These reviews, abundant in user experiences and perceptions, serve as a valuable asset for sentiment analysis. However, the sheer volume of these reviews poses a challenge for manual analysis, thereby necessitating the deployment of automated sentiment analysis methods. We utilize the Bernoulli Naive Bayes (BNB) and Support Vector Machine (SVM) models, selected for its simplicity, efficiency, and capability to handle large datasets with high dimensionality, characteristic of text data such as user reviews. Based on the analysis and conclusions derived from the research inquiries in this study, it is evident that a significant proportion of customer reviews, classifying 42.39% (or 142 reviews) as positive, 45.07% (or 151 reviews) as negative, and 12.54 % (or 42 reviews) as Neutral sentiments regarding the Desa Wisata Nusantara Application on the Google Play Store. We found that the Bernoulli Naive Bayes model had better accuracy than the Support Vector Machine. The accuracy level of the Bernoulli Naive Bayes model is 73 % and the Support Vector Machine is 70%. We present the outcomes of our analysis, discuss their implications, and propose directions for future research. Our research signifies a promising advancement towards harnessing machine learning for sentiment analysis, setting the stage for future investigations in this intriguing field.