Research on Detecting Credit Card Fraud Through Machine Learning Methods
Shunning Dai · Atlantis Highlights in Computer Sciences/Atlantis highlights in computer sciences · 2023
The heavy use of credit cards inevitably leads to the escalation of fraud technology and a surge in fraudulent behavior.Machine learning, a multi-interdisciplinary discipline with numerous algorithms, can effectively detect and prevent financial fraud.This study focuses on several common machine learning methods applied to fraud detection and then evaluates how they perform on real data, including Bagging, Random Forest, Decision Tree, and AdaBoost.However, the proportion of fraudulent transactions in real transaction data is extremely unbalanced.SMOTE can determine the data imbalance problem, while confusion matrices visualize the classification results of different classes.The experiment results reveal that Random Forest performs best for both unbalanced and balanced data.It indicates that random forest is better for detecting fraudulent transactions.