A Dynamic Resampling Method for Credit Card Fraud Detection in Imbalance Classification

The inconsistency of classes in the dataset is one of the primary challenges in data mining and machine learning, since it tends to forecast predictive results towards a specific class. In such instances, the synthetic minority oversampling technique (SMOTE) is the most commonly utilized resampling method, despite its class mixture and other limitations. To address class unbalanced difficulties in binary classification, in this research we present a dominant SMOTE oversampling method. The proposed method is compared to the standard SMOTE, Out Layer SMOTE, ADYASAN that belongs to the data level solution of the imbalanced dataset. In terms of precision, accuracy, recall, f1-score, true positive rate(TPR) and false positive rate(FTR), the experiment is validated using the credit card imbalanced dataset. The experiment result shows that the proposed method has achieved better results of 99% precision, 98% accuracy, 98% recall, 98% f1-score, 87% TPR and 13% FTR than traditional methods.