PENERAPAN DAN PERBANDINGAN ALGORITMA NAÏVE BAYES DAN K-NEAREST NEIGHBOR DALAM ANALISIS SENTIMEN TERHADAP KEPUASAN PENGGUNA APLIKASI FLO

Abstract

Digital technology is developing rapidly and has a wide and significant impact on the health sector, including through the presence of health monitoring applications such as the Flo application. This application is designed to help women track their menstrual cycles, fertile periods, and pregnancy. As an application that is personal and used routinely, user satisfaction is an important factor that determines the quality and sustainability of services. Sentiment analysis is needed to explore user views and preferences for this application. This study aims to analyze 18215 user reviews of the Flo application from the Google Play Store to classify sentiment, using web scraping techniques as a data retrieval method. Naïve Bayes and K-Nearest Neighbor are used as classification algorithms in data analysis. Data are analyzed through several stages, namely preprocessing, sentiment classification, model evaluation, and interpretation of results. The results showed that 93.1% of reviews were positive and 6.9% of reviews were negative. In terms of performance, the Naïve Bayes algorithm showed the best results with an accuracy value of 99%, precision 100%, recall 98%, and f-measure 99%, and without False Positive errors. Meanwhile, the K-Nearest Neighbor algorithm obtained an accuracy of 95%, precision of 97%, recall of 90%, and f-measure of 93%. The results of the study showed that the Naïve Bayes algorithm was more effective in analyzing the sentiment of Flo application user reviews.