Influence of Optimizing XGBoost to handle Class Imbalance in Credit Card Fraud Detection

XGBoost is one of the popular machine learning models used in the domains like fraud detection as well as to tackle the class imbalance that creates overfitting if not handled properly. Digital transactions are encouraged by financial institutions to maintain data integrity. Credit card payment is one of the prevalent means of transactions carried out in both online and offline purchases. Consequently, the risk of fraudulent activities are increased during these financial transactions. This creates awareness among the researchers about the need for an efficient method to detect fraud accomplishments. This paper proposes an optimized XGBoost (OXGBoost) approach to handle class imbalance in the datasets without using resampling techniques. In this proposed approach, RandomizedSearchCV hyperparameter optimization technique is applied to find the optimal parameters of XGBoost. The data sampling techniques are integrated with XGBoost to increase the efficiency of the model. The experiment was performed based on the two realworld credit card datasets. The findings of the experiment proved that the integration of data sampling does not have an impact on the efficiency of XGBoost. Based on the comparison, the proposed approach has outperformed the higher accuracy.