Predicting Credit Card Defaults with Machine Learning

Detecting credit card fraud is of paramount importance to financial institutions and credit card companies, as it plays a vital role in preventing monetary losses due to fraudulent transactions. The application of machine learning algorithms has become prevalent in this domain due to their capacity to analyze large datasets and identify patterns. In this specific study, we performed a comparative analysis of five machine learning algorithms, namely Logistic Regression, Decision Tree Classifier, Gaussian Naive Bayes, Random Forest Classifier, and Extra Trees Classifier, with the aim of detecting credit card fraud. The dataset employed in this study was characterized by its imbalanced nature, necessitating the use of the SMOTEENN algorithm to achieve balance in the dataset. To evaluate and compare the performance of the algorithms, various evaluation metrics were used, including precision, recall, F1-score, and the average precision-recall (AUPRC) curve. The outcomes indicate that the Extra Trees Classifier demonstrated superior performance across all metrics, including precision, recall, F1-score, and AUPRC. These results suggest that the Extra Trees Classifier exhibits significant potential as an effective instrument for detecting credit card fraud in real-world situations.