Analisis Sentimen Ulasan Pengguna BRImo Terhadap Pembaruan Fitur Aplikasi Menggunakan Naive Bayes Dengan Seleksi Fitur Chi-Square

Abstrak

Background: The BRImo app is a mobile banking service from Bank Rakyat Indonesia (BRI) with a large number of users. User reviews on Google Play Store are an important source of data for understanding user perceptions and satisfaction levels, but the number and diversity of review texts require an appropriate sentiment analysis method. Objective: This study aims to evaluate the performance of the Naive Bayes algorithm in classifying BRImo reviews by sentiment and the effect of Chi-Square feature selection. Methods: The research method includes text data preprocessing consisting of cleaning, case folding, tokenizing, normalization, filtering, and stemming. Next, feature weighting is performed using TF-IDF and feature selection using the Chi-Square method, followed by sentiment classification using the Naive Bayes algorithm. Model evaluation is performed using a confusion matrix with accuracy, precision, recall, and F1-score metrics. Results: The results show that the sentiment classification model achieved an accuracy of 95%, precision of 98%, recall of 96%, and an F1-score of 97%. The high recall value indicates the model's excellent ability to detect positive sentiment in user reviews. Conclusion: The Naive Bayes algorithm with Chi-Square feature selection is effective in analyzing BRImo application review sentiment and can be used as a basis for evaluating application development, but its performance is still limited in detecting negative sentiment due to the quantity of sentiment data.

Identifier

• DOI: https://doi.org/10.32665/statkom.v4i2.5194
• OpenAlex: https://openalex.org/W7118083618