Sentiment analysis of user comments on the shopeepay feature in the shopee application: Evaluation of accuracy with k-nearest neighbors (KNN) algorithm

Lestari, Fitri Duwi
Prasetiyo, Budi

Journal of Student Research Exploration
Vol. 3 No. 1 (2025): January 2025

This research analyzes Shopeepay user reviews on the Shopee app using the K-Nearest Neighbor (KNN) algorithm with TF-IDF weighting and a Cosine Similarity matrix. Data was collected through web scraping 500 reviews from the Google PlayStore and labelled into positive, neutral, and negative sentiments. The process includes literature study, data collection, labelling, text preprocessing, word weighting, and sentiment classification using KNN. Results show an accuracy range of 86%-91%, with Precision, Recall, and F1-Score as evaluation metrics. The findings indicate that convenience, trust, and risk significantly affect users' interest in Shopeepay, especially during the Covid-19 pandemic. A Word Cloud was also used to visualize common terms in the reviews, providing insights for Shopee to enhance Shopeepay based on user feedback.