ELIGIBILITY CLASSIFICATION FOR BANK LOANS USING PROBABILISTIC NEURAL NETWORK, K-NEAREST NEIGHBOR, AND NAIVE BAYES CLASSIFIER ALGORITHM Lena Oktavianis, Mustakim, Rice Novita, Inggih Permana
Universitas Islam Negeri Sultan Syarif Kasim Riau
Abstract
Determination of bank creditworthiness is very influential on the income of a bank. If the bank is not right in determining credit loans, then the credit can become problematic or bad which ultimately results in losses to the bank. Therefore, a classification method is needed to minimize the occurrence of non-performing loans. This research applies classification to customer data of PT. BPR Fianka Rezalina Fatma in 2018-2020 using classification methods, namely the Probabilistic Neural Network (PNN), K-Nearest Neighbor (KNN), and Naive Bayes Classifier (NBC) algorithms. This study also applies data sharing techniques to share training and testing data using hold-out techniques. This study found that the PNN algorithm has the highest accuracy, 98.25% with a spread of 0.001.
