Avriliana, Disty (2025) Implementasi Algoritma NaÏve Bayes Classifier Untuk Analisis Sentimen Ulasan Aplikasi Tokopedia Pada Situs Google Playstore. Other thesis, Universitas Islam Riau.

Text 203510539.pdf - Submitted Version Restricted to Registered users only Download (1MB) | Request a copy

Abstract

The proliferation of e-commerce in Indonesia has led to an increased emphasis on understanding user opinions to enhance services. Tokopedia, a prominent e- commerce platform, receives a substantial volume of reviews on Google Play Store. These reviews can be systematically analyzed to ascertain customer sentiment. Given the sheer number of reviews, this study employs the Naïve Bayes Classifier to automate the classification of user sentiment. A total of 7,000 reviews were collected using web scraping techniques between January 2020 and January 2025. The data was labeled with few-shot prompting using ChatGPT and processed through preprocessing and TF-IDF weighting stages before being classified with Naïve Bayes Classifier. The model's performance was benchmarked against Logistic Regression, SVM, and Random Forest algorithms, with metrics including accuracy, precision, recall, and F1-score.The evaluation revealed that the Naïve Bayes Classifier attained 76% accuracy, while the Logistic Regression algorithm achieved the highest accuracy of 93%, followed by SVM (86%) and Random Forest (85%).Cross-validation analysis indicated that the Naïve Bayes Classifier exhibited greater variability in accuracy compared to the other methods. The present study posits that Logistic Regression is a more optimal approach for sentiment analysis of Tokopedia reviews, while the Naïve Bayes Classifier remains an efficient alternative for large-scale text processing.

Actions (login required)

View Item