A Heterogeneous Ensemble Learning Model Based on Data Distribution for Credit Card Fraud Detection

Credit card fraud detection (CCFD) is important for protecting the cardholder’s property and reputation of banks. Class imbalance in credit transaction data a primary factor affecting classification performance current models. However, prior approaches are aimed at improving prediction accuracy minority class samples (fraudulent transactions), but this usually leads to significant drop model’s predictive majority (legal which greatly increases investigation cost In paper, we propose heterogeneous ensemble learning model based on distribution (HELMDD) deal with imbalanced CCFD. We validate effectiveness HELMDD two real datasets. The experimental results demonstrate that compared state-of-the-art models, has best comprehensive performance. not only achieves good recall rates both also savings rate banks 0.8623 0.6696, respectively.