<![CDATA[This research is motivated by the rapid growth of financial technology in Indonesia, where the DANA application has become the most popular digital wallet (e-wallet) with over 200 million registered users . The high usage of this application results in an abundance of reviews on the Google Play Store, representing both customer satisfaction and complaints . The problem addressed in this research is how to automatically process these textual reviews and determine the best classification method among the three tested Machine Learning algorithms . This research aims to analyze and compare the accuracy performance of Naive Bayes, Decision Tree, and Support Vector Machine (SVM) algorithms in classifying user sentiment . The method used in this research is a computational quantitative approach, utilizing a secondary data collection technique consisting of 50,000 reviews from the Google Play Store via Kaggle . The analysis process was conducted by applying five stages of text preprocessing, feature weighting using Term Frequency-Inverse Document Frequency (TF-IDF), handling data imbalance using the Synthetic Minority Over-sampling Technique (SMOTE), data splitting (80% training data and 20% testing data), and model evaluation using a Confusion Matrix . The results showed that the Naive Bayes algorithm had the most superior performance with an accuracy rate of 80%, followed by Decision Tree and Support Vector Machine (SVM), each obtaining an accuracy of 78% . Therefore, it can be concluded that the Naive Bayes algorithm is the most optimal and stable method for conducting sentiment analysis classification on e-wallet application review text data after the class distribution is equalized.]]>
Copyrights © 2026
