Machine Learning Predictive Model for direct card fraud in Skewed Data Detection

Abstract:

The problem of fraudulent payment card transactions is becoming more and more serious in the era of remote services and e-commerce. Analyzing thousands and millions of data and finding invalid transactions among them is a problem. Machine learning techniques are dedicated to processing extensive data. The selected mechanisms are successfully used in detecting fraudulent payment transactions. The paper presents research using Decision Tree Classification Model and Random Forest Model with two estimator values 50 and 100. Confusion Matrix and K-Fold Cross Validation Metric were used as metrics.