Experimental Report on the Comparison Between Generalized Mapping and Markov Chain in E-commerce Fraud Detection

Abstract

This experiment focuses on the field of e-commerce fraud detection, aiming to compare the performance of the generalized mapping rule system and the Markov chain state transition system. By programming with Python, 5000 simulated order data with 9 features were generated, and the two systems were used for processing respectively. The experimental results were analyzed from aspects such as time complexity and accuracy to provide data support for the selection of e-commerce risk control strategies. It was found that the generalized mapping has a relatively high accuracy but insufficient recall for fraud orders, while the Markov chain has high processing efficiency but serious misdetection.