<p>This study mainly analyzes the current fierce competition in the e-commerce market, as well as the high and unstable churn rates of e-commerce platforms and customers. A new model based on random forest algorithm and extreme gradient boosting algorithm has been built this time. This model analyzes the different situations and reasons of customer churn by building a regression decision tree module, and then adds an extreme gradient enhancement algorithm to enhance the model’s predictive ability and accuracy in predicting customer churn. The results indicated that the majority of customer churn on e-commerce platforms was related to the quality of products and after-sales service. The better the after-sales service and products, the less customer churn occurred on e-commerce platforms. The research model had better model performance, with a root mean square error of about 1.4% lower than the highest model, a highest accuracy of 94.6%, and a recall rate of 7.5% higher than the lowest model. The training time of the research method was stable at around 300&#xa0;s, with less training time and higher prediction efficiency. Therefore, the new model used can better predict the current customer churn situation on e-commerce platforms.</p>