Indonesia Critical Sentiment Analysis of Tokopedia Electronic Products Using SVM-Logistic & TF-IDF Ensemble Methods

Muhammad Zainottah Zainottah, Rengga Saputra Rengga, Yustian Servanda Yustian, Isa Rosita IsaNon-Scopus15 Juni 2025Afiliasi: ID

Abstrak

This research aims to analyze customer review sentiment for electronic products on Tokopedia using a Support Vector Machine (SVM) classification method with Term Frequency-Inverse Document Frequency (TF-IDF) based features, enhanced by an ensemble approach with logistic regression. Utilizing a Tokopedia review dataset from 2023, this study seeks to identify critical sentiments embedded in customer reviews, which can provide valuable insights for sellers and the platform. The methodology involves comprehensive textual data preprocessing, feature extraction using TF-IDF for vector representation, and the implementation of an SVM-Logistic ensemble model via a stacking strategy. The results indicate that the SVM-Logistic ensemble model can classify review sentiments with high accuracy and superior performance metrics, effectively distinguishing between positive, negative, and neutral sentiments. These findings highlight the significant potential of machine learning methods in automatically understanding customer feedback, which is crucial for continuous improvement in product and service quality on e-commerce platforms, and for supporting more strategic business decisions.

Identifier

Scroll to Top