Predicting fraud in credit card transactions

The exponential growth of internet-based services has led to an increase in credit card fraud, posing significant financial risks to users and institutions. This study shows the application of supervised machine learning algorithms—specifically Decision Tree and Random Forest classifiers—for effective detection and prediction of fraudulent credit card transactions. Using a large, simulated dataset of 555,719 transactions with both legitimate and fraudulent cases, we addressed the severe class imbalance through an under sampling technique. Our results demonstrate that the Random Forest model outperforms the Decision Tree, achieving an accuracy of 95.80%, sensitivity of 95.80%, precision of 99.58%, and F1 score of 97.49%.