Analisis Sentiment Review Produk Tokopedia Menggunakan Ensemble Learning

Abstract

The rapid development of e-commerce in Indonesia has encouraged product review analysis to improve service quality and facilitate consumer decision-making. One of the popular e-commerce platforms, Tokopedia, relies on product reviews as a means to build trust and influence purchasing decisions. However, with the large number of reviews available, manual analysis becomes inefficient and error-prone. Therefore, this study aims to apply sentiment analysis to product reviews on Tokopedia using ensemble learning-based machine learning algorithms, namely Decision Tree, Random Forest, AdaBoost, and XGBoost. This study began by taking a product review dataset from the Rexus store on Tokopedia, consisting of 470 reviews which were then processed through pre-processing stages such as text cleaning, stopword removal, stemming, labeling with VADER, and feature extraction using TF-IDF. In terms of accuracy, the best of the four models is Random Forest and AdaBoost with an accuracy of 81%, then XGBoost with an accuracy of 80%, and finally Decision Tree with an accuracy of 79%.