ANALISIS SENTIMEN OPINI PENGGUNA JASA PENGIRIMAN JNE MENGGUNAKAN METODE NAÏVE BAYES CLASSIFIER DAN K-NEAREST NEIGHBORS

JNE's high-quality service will provide optimal satisfaction to users, ensuring they feel valued and have a reliable and efficient delivery experience. To provide optimal service, this research explores in-depth user sentiment analysis of freight forwarding applications in Indonesia. The purpose of the study is to analyze user sentiment towards the My JNE app, which is one of the leading freight forwarding apps in Indonesia. This research uses user review data from Google Play Store collected from 2018 to 2024. The review sentiment is categorized into positive, neutral, and negative using the VADER analysis tool. VADER (Valence Aware Dictionary and sEntiment Reasoner) is a lexicon and rule-based sentiment analysis tool specifically designed to detect sentiment in social media text. After the data reduction process, neutral sentiment classes were removed to focus the analysis on two main categories: positive and negative. Of the total 5,000 review samples analyzed, it was found that 35.78% belonged to the positive category and 64.21% to the negative category. The classification methods used in this study are Naïve Bayes and K-Nearest Neighbors (KNN). The analysis results show that the Naïve Bayes model has an accuracy of 81.64%, while K-Nearest Neighbors (KNN) has an accuracy of 76.25%. This accuracy test confirms that the KNN model is more effective in classifying user sentiment compared to Naïve Bayes. The results of this study provide important insights into user perceptions of the My JNE application, which can be used as a basis for improving service quality in the future. This research suggests that My JNE focus on improving features that often receive negative reviews to increase user satisfaction.