Research on Credit Card Fraud Detection Model Based on Class Weighted Support Vector Machine

Abstract

To deal with credit card fraud, a detection model based on Class Weighted Support Vector Machine was established. Due to large-scale and high dimensions of data, Principal Component Analysis (PCA) was adopted firstly to screen out the main factors from a great deal of indicative attributes in order to reduce the training dimension of SVM effectively. Then according to the characteristics of credit card transactions data which are imbalance, an improved SVM--Imbalance Class Weighted SVM (ICW-SVM) was adopted. With the application and verification in real dataset from bank, it is demonstrated that this model is more suitable for solving credit card fraud detection problem with higher precision and effective than others.