Efficient Fraud Detection Classification: Class Imbalance and Attribute Correlations
Difei Liu, Ruiyi Sun, Haoyang Ren · The Frontiers of Society Science and Technology · 2020
Fraud detection is a specifically important issue to protect cardholders’ information from being stolen by fraudsters. By choosing proper algorithms and analyzing behavioural information of cardholders and banks, we can significantly reduce the probability of transactions being illegally manipulated. In response to possible problems in fraud analysis, this article will focus especially on tackling class imbalance problems and finding attribute correlations. Two FraudDetection datasets on Kaggle will be used to build classifiers and ananlyze the impact of different data processing techniques. Through this process, we realized recent findings of fraud detection, we got to know more about different data processing methods, and we implemented distinct types of classifiers. We confirmed the significance of class imbalance tackling and attribute correlations analyzing.