Detection of credit card fraud |
Author(s): |
| Goutam Patil , East West Institute of Technology; Anjana H S, East West Institute of Technology; Jayanth G, East West Institute of Technology; Manjunath M, East West Institute of Technology; Mayur Ravikumar Heggade, East West Institute of Technology |
Keywords: |
| Credit Card, Fraud System |
Abstract |
|
Credit card fraud causes significant financial losses, demanding robust fraud detection systems. This study focuses on addressing challenges such as data imbalance and evolving fraudulent behavior using machine learning techniques. The dataset consists of 284,807 transactions over two days, where fraud accounts for only 0.172%. We use five machine learning algorithms—Logistic Regression, K-Nearest Neighbors, Decision Tree, Random Forest, and XGBoost—alongside six resampling techniques: Random Oversampling, SMOTE, Random Undersampling, Tomek Links, Cluster Centroids, and SMOTE + Tomek Links. XGBoost outperformed all other models across seven scenarios using the ROC-AUC as the evaluation metric. The findings emphasize oversampling techniques, particularly SMOTE and SMOTE + Tomek Links, in improving model performance. |
Other Details |
|
Paper ID: IJSRDV12I100041 Published in: Volume : 12, Issue : 10 Publication Date: 01/01/2025 Page(s): 38-39 |
Article Preview |
|
|
|
|
