Klasifikasi Naive Bayes pada Data Tidak Seimbang untuk Kasus Prediksi Resiko Kredit Debitur Kartu Kredit.

Abstract

Banking business in Indonesia is still dominated by the credit business field. Most of the bank's revenue comes from this business field. Unfortunately, credit risk can cause problems in loans which can reduce the bank’s revenue. This research uses a Naive Bayes classification analysis on the imbalanced data for the predictions of debtor’s credit risk that are able to classify the future debtor into the following two categories: good or bad. Sampling strategy is used to overcome the problems of imbalanced data. Duplication oversampling, random oversampling, random undersampling, and cluster undersampling are chosen as the methods. It is found that the random oversampling method shows the best value after sampling strategy is conducted with an f-measure of 83.30%.