Background: The main purpose of this thesis is in electronic commerce, reliable anomaly detection systems are essential for maintaining security and improving user experiences, especially in the electronics industry. With the goal of filling in the gaps in the current anomaly detection methods, this study examines the efficacy of SVM as a fundamental algorithmic framework for anomaly detection in retail electronics online. The goal of the research is to better understand user interfaces and security protocols one-commerce platforms by spotting anomalies within user behavioral events.

Objectives: To evaluate the effectiveness of SVM in identifying anomalies in user activity patterns, a rigorous experimental design comprising feature extraction, preprocessing, and model evaluation is used in the technique.

Methods: The study establishes the foundation for analysis and model creation by utilizing data from the REES46 platform, which records a broad range of user interactions over an extended period of time. Utilizing this extensive dataset, the study explores the subtle aspects of user behavior and offers insights into SVM algorithm-based anomaly detection methods. The methodology’s rigorous data preprocessing and feature extraction ensured the dataset’s integrity, contributing to the model’s ef-fectiveness. Metrics including precision, recall, and F1-score were used to train and assess the SVM model after a thorough normalization of the dataset using Stan-dard Scaler. With an F1 score, a precision, and a recall. The model’s accuracy was further confirmed by a low Mean Squared Error (MSE), Prediction scatter plots and other visualizations.

Results: The findings highlight the considerable potential of SVM-based anomaly detection systems to improve user experiences and strengthen security protocols in online retail settings. higher scores for the classification metrics. The model performed well, obtaining an F1 score of 0.93, a precision of 99per, and a recall of 1.00, demonstrating its great accuracy in identifying anomalies while reducing false positives and negatives respectively. Minimal difference between expected probabilities and actual values was indicated by a low Mean Squared Error (MSE), which further supported the model’s precision. The results show that SVM could significantly improve security and user experience by accurately identifying anomalous user behaviors that may indicate fraud or other dangerous behaviors.

Conclusion: In conclusion, by showing the efficiency of SVM algorithms, this study advances anomaly detection in online electronics shopping. Consistent with the initial goals mentioned in the introduction, the study tackles important e-commerce problems and provides insightful information about how to apply innovative machine learning methods to improve security and user experiences. The results open up new avenues for anomaly detection system research and innovation, which will enhance online retail environments.