Securing e-commerce platforms through a support vector machine approach using multifactor authentication to enhance security and compliance

The system explosion on the e-commerce sites has given a huge thrust to the volume of cybersecurity attacks, such as fraud transactions, identity theft, and non-compliance issues. This paper proposes an integrated security model that implements Principal Component Analysis (PCA) for fraud detection and Risk-Adaptive Multifactor Authentication (MFA) for secure user authentication. The system uses Synthetic Minority Over-sampling Technique (SMOTE) for accurate class imbalance handling and validates the classifier’s performance. PCA is applied to decrease feature dimensionality through transformation of the transactional features into small, uncorrelated components, thus enhancing feature separability and classification efficacy before detecting frauds. Testing of the model takes place using a real-world dataset of cardholder transactions in Europe, which includes principal component features derived from transaction metadata. Experimental results show the combined Support Vector Machine (SVM)–SMOTE method supported by good Area Under the Curve (AUC) and precision–recall curve metrics. Moreover, the application of Risk-Adaptive MFA control enhances security by making every transaction pass through multiple levels of identity verification. The proposed architecture combines SVM-based fraud detection with MFA that adapts according to the level of risk. It provides an AUC of 0.9981 and an F1-Score of 0.984, thereby demonstrating excellent predictive power in terms of its evaluation. Robust functioning can be ensured despite changing levels of transactions and behaviors. Real-time fraud detection and transaction processing can be facilitated through this framework. The paper highlights that combining machine learning algorithms and secure authentication processes can be used effectively to create an efficient and secure e-commerce system. The alignment with Trends in Social Computing for Customer Relationship Management can be accomplished by incorporating intelligent fraud detection mechanisms and adaptive authentication mechanisms in the e-commerce systems.