<![CDATA[The credit procedure is the provision of credit on the basis of the bank's belief in the ability and ability of the customer to repay. In this study, the customer data tested was divided into 3 classes, namely 39 LOW customers, 84 MEDIUM customers, and 11 HIGH customers. Unbalanced datasets are pre-processed using the Synthetic Minority Over-sampling (SMOTE) technique. Classification methods such as Random Forest and Support Vector Machine will test cooperative customer data. The data is tested based on the attribute that has the highest matching value using the Particle Swarm Optimization method, this test is also optimized with the Adaboost method to increase its accuracy. The results of tests carried out using the Random Forest method obtained an accuracy of 89.05% and the Support Vector Machine algorithm obtained an accuracy of 81.75%. Meanwhile, testing with two methods optimized with Adaboost showed an increase in accuracy with Random Forest getting 91.24% accuracy and Support Vector Machine getting 82.48% accuracy. The highest accuracy in the cooperative customer data classification test was obtained from the Random Forest algorithm which was optimized with Adaboost at 91.24%]]>
Copyrights © 2023
