Optimization Accuracy of Fraud Detection in E-Commerce using ML Technique

Abstract

The rapid growth of e-commerce platforms has significantly increased the risk of fraudulent transactions, leading to substantial financial losses and reduced customer trust. This study focuses on optimizing the accuracy of fraud detection systems using advanced machine learning (ML) techniques. A comprehensive framework is proposed that integrates data preprocessing, feature engineering, class imbalance handling, and model optimization to enhance detection performance.

Initially, transactional datasets are cleaned and transformed using normalization and encoding techniques. Due to the highly imbalanced nature of fraud datasets, Various ML classifiers, including Random Forest, Support Vector Machine (SVM), Gradient Boosting, and Extreme Gradient Boosting (XGBoost), are implemented and evaluated. To further improve accuracy, hyperparameter optimization techniques such as Grid Search and Bayesian Optimization are utilized. The performance of the proposed models is assessed using evaluation metrics such as accuracy, precision and recall. Experimental results demonstrate that optimized ensemble models outperform traditional approaches, achieving higher detection accuracy and reduced false positives. The proposed system provides a robust and scalable solution for real-time fraud detection in e-commerce environments, enhancing transaction security and customer confidence. Future work may involve integrating deep learning techniques and real-time adaptive learning mechanisms to further improve system performance.