Fraud credit card transaction detection using hybrid multilayer perceptron-random forest method

Credit card fraud is a leading crime with rapid growth in the world. This is due to credit cards being one of the most popular payment options worldwide. To address this problem, there needs to be a robust and efficient method to accurately identify fraudulent transactions. This study aims to investigate the performance of a hybrid method that combines Multilayer Perceptron (MLP) as a feature extractor and a Random Forest (RF) classifier for detecting fraudulent credit card transactions. The MLP is used to capture complex patterns in the transaction data, while the RF classifier is used to make robust and accurate predictions. The performance of the proposed model was compared with standalone MLP and RF using several evaluation metrics. The proposed method achieved the best performance among other methods, with an accuracy of 99.949%, precision of 87.097%, recall of 82.653%, and F1-score of 84.817%. This result shows the ability of the proposed method by combining the strengths of MLP as a feature extractor and RF as a classifier, offering an effective and robust method for fraud detection. This research shows the potential of hybrid methods in addressing financial challenges and provides further advancement in fraud detection systems.