<![CDATA[Customer satisfaction is a crucial element that plays a significant role in the sustainability of businesses in the e-commerce sector. Reviews provided by consumers serve as an important source of information to assess how satisfied they are with the products they purchased. This study aims to evaluate customer satisfaction levels using product review data through two classification methods: Multinomial Naive Bayes and Logistic Regression. The data used comes from a real Indonesian-language dataset that includes review texts and buyer ratings. The research process consists of several stages, starting from text preprocessing, feature extraction using the TF-IDF method, satisfaction label grouping, model training, and evaluation using metrics such as accuracy, precision, recall, F1-score, and confusion matrix. The findings of this study indicate that both methods can predict customer satisfaction with competitive accuracy. Logistic Regression demonstrates more consistent results compared to Naive Bayes in the context of Indonesian-language text. These results can be utilized by e-commerce companies to monitor product quality and continuously improve services for consumers.]]>
Copyrights © 2026
