Advanced machine learning techniques for credit card fraud detection
Abha Kumari, Ishita Pandey, Vanshika Gupta, Yash Veer Singh · 2025
Authors This project addresses the critical challenge of identifying fraudulent transactions inelectronic commerce. The project begins with the acquisition of a dataset, followed by rigorous data preprocessing and feature scaling to optimize thedataset for analysis. Three distinct ML algorithms—Random Forest Algorithm, Logistic Regression Algorithm, and Decision Tree—are employed to train and test the model, with performance evaluation based on metric measures like Precision, Recall-Score, and F-Score. Acknowledging the imbalances inherent in fraud detection datasets, the project implements Under Sampling andOversampling techniques to enhance model robustness. The results of these techniques are thoroughly evaluated, leading to the identification of the most effective approach. Upon comparison, the experimental outcomes reveal that the Random Forest Classifier outperforms other algorithms, achieving an exceptional accuracy of 99.99%. Logistic Regression and Decision Tree Classifier also demonstrates commendable accuracy rates with 94.45% and 99.79%.