Perbandingan Kinerja Algoritma Naive Bayes dan K-Nearest Neighbor dalam Menganalisis Sentimen Pengguna Game Free Fire

Abstract

Free Fire is one of the most popular online games in Indonesia, yet it continues to receive a wide range of user reviews regarding gameplay experiences. These reviews reflect diverse user perceptions, including both praise and criticism, making sentiment analysis essential to understanding user satisfaction. This study aims to classify user sentiments toward Free Fire using a combined dataset collected from the Google Play Store and App Store, and to compare the performance of two text classification algorithms: Naive Bayes and K-Nearest Neighbor (KNN). The data were collected using web scraping techniques and manually labeled by expert validators. Text preprocessing involved cleansing, tokenizing, stopword removal, and stemming, followed by term weighting using the Term Frequency-Inverse Document Frequency (TF-IDF) method. The experimental results show that the Naive Bayes algorithm achieved the highest accuracy of 72.78%, while the KNN algorithm recorded a maximum accuracy of 45.91%. Based on these findings, Naive Bayes is proven to be more effective in classifying user sentiments related to Free Fire. The results of this study are expected to provide constructive insights for developers to improve the quality and user experience of the game.