Perbandingan Algoritme C4.5 dan CART pada Data Tidak Seimbang untuk Kasus Prediksi Risiko Kredit Debitur Kartu Kredit

Abstract

Credit risk is the risk of customers inability regarding their debt payment obligations. Credit risk can lead to non-performing loans, thereby reducing bank earnings. Various techniques can model the revenue of a debtor. One of them is the decision tree. There are several decision tree algorithms that can be used as classifiers, for example: C4.5 and CART. However, the classification results may be inaccurate due to the imbalanced in the used data. One way that can be used to address this problem is to include methods of sampling strategies: for examples over-sampling and under-sampling. This research found that the implementation of sampling strategy on the use of C4.5 and CART can improve the average accuracy, precision, recall, and F-measure.