SUMMARY
Credit scoring in financial institutions/banks is very important in determining whether it is feasible or not to receive borrowed funds. Therefore, credit scoring is the most important part of the company and has a great impact on the company's profits. Therefore, there are very strict regulations in the credit scoring process. This credit evaluation uses customer data collected by filling out forms, customer financial history data, and various other data sources. The Optimal Binning Algorithm is an industry-standard method used by data scientists to build credit scoring models. However, data scientists need to analyze each variable or feature individually to understand how the pattern of each variable affects the goal. When there are hundreds or even thousands of variables, it will take a lot of time, From a business perspective, it usually requires a solution that is as fast and effective as possible. To solve this problem, we propose the implementation of evolutionary algorithms to find the best solution of variable value constraints to form optimal binning. Genetic algorithms as a branch of adaptive evolutionary algorithms commonly used to solve value searches in optimization problems. Through this research, genetic algorithms can solve the problem of determining the limitations in the formation of optimal binning ariables very well. The results of the comparison between the classifier methods provide evidence that Logistic Regression can consistently have stable performance. In addition, it can maintain the performance of the predictive model despite changes in trends in the test data, while other methods get a decrease in performance results. In the Logistics Regression t test data, he received an AUC of 0.71