This project aims to evaluate the efficacy of machine learning and deep learning algorithms in predicting credit card defaults and to discover the optimal balancing strategies that aid in improving the accuracy of the prediction. Additionally, the study will identify the optimal balancing approaches that help in predicting credit card defaults.
Which is the most effective balancing technique that helps in improvising the accuracy of the machine learning and deep learning models?
• To implement machine learning algorithms such as decision tree, Adaboost, XGBoost, and SVM and deep learning algorithms such as MLP classifier to detect credit card fraud with and without balancing the data. • To implement balancing techniques such as oversampling such as random over sampling, ADASYN and SMOTE, and undersampling such as random undersampling technique. • To evaluate the performance of balancing techniques based on accuracy, precision, recall and F1 score and to compare the performance of machine learning and deep learning algorithms in credit card fraud detection.