Advanced Credit Scoring with Naive Bayes Algorithm: Improving Accuracy and Reliability in Financial Risk Assessment

This study develops a credit application recommendation system based on the Naive Bayes method to improve accuracy and reliability in financial risk assessment. Using the CRISP-DM framework, the research process starts with understanding the business needs to implement a web-based system. The Naive Bayes algorithm was chosen because of its ability to handle binary data classification and generate reliable predictions even with limited training data. This study combines feature selection and unbalanced data handling techniques to improve model performance. The evaluation results showed that the system achieved an accuracy of 70.50%, with a precision of 92.16%, a recall of 64.57%, and an F1-score of 75.83%. This system is implemented as a web-based application to help financial institutions make credit decisions quickly and accurately. These findings significantly contribute to developing a data-based classification system for the banking sector, especially in reducing the risk of bad loans and improving decision-making efficiency