Penerapan Algoritma Decision Tree C4.5 Untuk Deteksi Fraud Pada Kartu Kredit dengan Oversampling Synthetic Minority Technique (SMOTE)

Ludgerus Darell Perwara (Fakultas Ilmu Komputer, Universitas Brawijaya)
Fitra Abdurrachman Bachtiar (Fakultas Ilmu Komputer, Universitas Brawijaya)
Indriati Indriati (Fakultas Ilmu Komputer, Universitas Brawijaya)

Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer
Vol 4 No 8 (2020): Agustus 2020

In today's technological developments, credit cards are seen as an easy and practical way of making transactions, because apart from being easy to use transactions with credit cards only require a few requirements. However, the increasing use of credit cards has resulted in criminal acts that harm both customers and banks. Data mining is seen as the right method to solve this problem, so this research will use the Decision Tree C4.5 method to detect fraud in credit card transactions. Because the occurrence of fraud on every transaction is rare and there are more normal transactions, this study will also add to the SMOTE oversampling method that can create synthetic fraud data with the aim of balancing class. The results of this study produce an accuracy value of 78%, a precision value of 89.65%, a recall value of 85.71% and an f-measure of 87.64% with an N value of 500% in SMOTE and a depth value of the Decision Tree C4. 5 = 15. So, it can be concluded that the implementation of Decision Tree C4.5 in the case of detecting fraud in credit card transactions is best done by oversampling SMOTE.