Probabilistic Learning Models for Secure E-Commerce Transaction Analysis

Abstract

The tremendous expansion of e-commerce requires well-developed security mechanisms of transactions coupled with cryptography systems. This paper explores how probabilistic learning algorithms can be used to promote fraud and transaction validation in e-commerce systems that are cryptographically secured. A study compared different machine learning models to analyze retail data, including Naïve Bayes, Bayes Network, and Logistic learning. The results showed that different models excelled in various metrics, with Logistic learning achieving 88.73% accuracy in one comparison and Naïve Bayes showing 88.93% accuracy in another. Naïve Bayes also performed well in other measures like precision, recall, and F-Measure. Overall, the study found that the Naïve Bayes model performed best across various performance measurements, suggesting it could be particularly useful for analyzing smart retail data. The fusion of probabilistic learning as well as cryptographic systems could be used in payment transaction security and detecting frauds.