IJAIEM

Vignesh , Dharshanala Nithin , Alineni Sai Chandu ,M. Hari Kumar

Abstract

Fraudulent credit card transactions pose a significant challenge in the finance sector, leading to substantial financial losses. Traditional rule-based techniques have proven inadequate in handling the multitude of variables associated with fraud detection. To mitigate these issues, this study employs advanced machine learning techniques to enhance the accuracy and efficiency of fraud detection systems. The project focuses on utilizing algorithms such as Random Forest and TabNet to categorize transactions as legitimate or fraudulent. The data preprocessing steps include cleaning and normalizing the input data to ensure consistency and accuracy. Key features are extracted and encoded to provide the most relevant information for the models. Model evaluation is performed using metrics such as precision, recall, and F1-score, and the results are visualized through confusion matrix heatmaps. The hybrid approach combining Random Forest and TabNet aims to provide a robust solution for