Abstract

This research intends to build an application that can analyze bank data and then determine the feasibility in terms of creditworthiness, to avoid non-performing loans in the future. The method used is a hybrid method that combines two Data Mining classification techniques namely Support Vector Machines (SVM) and K-Nearest Neighbors (KNN). SVM works by finding the optimal hyperplane and support vectors. Furthermore, the KNN will classify bank data based on identifying the support vectors. With 2000 training data and 103 testing data: cost parameter values = 0.1, gamma = 2, 1998 support vectors, then with K value = 16 the system gives 88.35% suitable data (91 data from 103). In conclusion, the application can work in terms of helping the credit analysts to recommend prospective customers who deserve loans.
Keywords – application; data mining; hybrid method; SVM-KNN