Data Characteristic Stability Based Random Forest Implementation of Credit Card Fraud Detection
Srinath Mugundhan, Pranesh Venkataramanan · 2022
The rapid growth of internet enabled technology enhanced the digital paying system in recent days. Financial Applications in online mode become much easier. People access the internet on the go and do Applications in a fraction of seconds. On the other hand digital environment allows the fraudulent are unauthorized attackers through loop holes available in the internet platform. This paper proposes a Data Characteristic Stability based feature selection by implementing the Random Forest algorithm for Credit Card Fraud Detection. After implementing the above optimized method for credit card fraud detection, the proposed detection engine result is compared against without applying Data Characteristic Stability using the following ML classifiers: Decision Tree (DT), Random Forest (RF), Logistic Regression(LR), XGBoost and etc. The Proposed solution is considered to be efficient an way because of the tendency of the drift happens after its deployed into production, where now we identify the indicative data drift during the Model development and helps to outperform in identifying the fraudulent cases way better than the traditional methods used in the market.