ANALISIS SENTIMEN PENGGUNA DOMPET DIGITAL MENGGUNAKAN ALGORITMA MULTIVARIAT BERNOULLI (STUDI KASUS: OVO DAN GOPAY)

This research aims to analyze user sentiment towards two leading digital wallets in Indonesia, OVO and GoPay, using the Multivariate Bernoulli algorithm, a member of the Naive Bayes family effective in text classification. Data was collected from 1,070 user reviews through questionnaires distributed on social media and discussion forums, resulting in 915 sentiment data points after data cleaning. The preprocessing process included cleansing, case folding, tokenizing, normalization, filtering, and stemming to prepare the data for sentiment analysis. The study employed the K-Fold Cross Validation technique with 5 folds to test the model and obtain average accuracy, precision, and recall. The results showed that the Multivariate Bernoulli model performed well with an average accuracy of 86.20% for OVO and 83.15% for GoPay, with very high recall values indicating the model's ability to detect positive sentiment. The confusion matrix from each fold demonstrated consistent ability to identify positive cases. This study's findings are expected to provide recommendations for the development of OVO and GoPay digital wallet services based on user sentiment analysis.