Detection of Credit Card Fraud with Machine Learning Methods and Resampling Techniques

Financial institutions in the form of banks provide facilities credit cards, but with development technology, fraud on card transactions is still common, so a system needed that can detect quickly and accurately. Therefore, this study aims to classify fraudulent transactions. The proposed method Ensemble Learning which will be tested using Boosting type 3 variations, namely XGBoost, Gradient Boosting, AdaBoost. Then, maximize performance model, dataset used optimized Synthetic Minority Oversampling Technique (SMOTE) function from Imblearn library data train handle imbalanced conditions. entitled "Credit Card Fraud Detection" total 284807 divided into two classes: Not Fraud. model received recall 92% where results increased by 10.37% compared previous Random Forest result 81.63%. This because use SMOTE greatly influences classification classes.