The development of online lending services in Indonesia has led to various responses from the public on social media, including complaints about billing methods and concerns about high interest rates. This study aims to compare the performance of Naive Bayes, Decision Tree, and K-Nearest Neighbors (KNN) algorithms. This type of research is quantitative, and the data used is 5,941 tweets through crawling techniques from X social media, followed by preprocessing, data labeling with a lexicon-based feature extraction using TF-IDF, and sentiment classification using the three algorithms. The evaluation stage uses a confusion matrix, which can calculate accuracy, precision, recall, and the F-1 score. The results show that the decision tree provides the most consistent performance with 69% accuracy due to its ability to recognize complex data patterns and understand relationships between features. Naive Bayes excels in negative sentiment classification with 68% accuracy, while KNN shows the lowest performance with 44% accuracy because it is not effective in handling high-dimensional text data. These results can be utilized by online loan service providers and regulators to build an accurate public opinion monitoring system in order to respond to issues of public concern and improve service quality on an ongoing basis.
Miranda et al. (Sun,) studied this question.