KOMPARASI ALGORITMA NAIVE BAYES DAN K-NEAREST NEIGHBOR PADA ANALISIS SENTIMEN TERHADAP ULASAN PENGGUNA APLIKASI TOKOPEDIA

Abstract

Tokopedia is one of the leading e-commerce platforms in Indonesia. The use of e-commerce platforms has increased rapidly in recent years. This is due to technological advances, increased internet access, and consumer behavior that prefers to shop online. In today's digital era, user reviews have an increasingly important role in shaping consumer perceptions of a product or service. The purpose of this research is to conduct sentiment analysis on application performance based on user reviews of the Tokopedia application. Researchers made the decision to use sentiment analysis because it is the most suitable method for processing data sets. From 1019 Tokopedia user reviews on the Play Store that were collected, 176 positive reviews and 843 negative reviews were obtained. Then, the data is classified using the Naive Bayes and K-Nearest Neighbor algorithms, then optimized using Particle Swarm Optimization. The results of the research conducted obtained an accuracy of 76.30% for the Naive Bayes accuracy value without feature selection, 74.09% for Naive Bayes results using feature selection. Then the accuracy value obtained for K-Nearest Neighbor without feature selection is 83.10%, and with feature selection is 83.53%. From the results obtained, the effect of using Particle Swarm Optimization selection features on the two algorithms does not have a big impact, there is an insignificant change in accuracy and AUC values which in the Naïve Bayes algorithm actually decreases