Comparison of Model in Predicting Customer Churn Based on Users' habits on E-Commerce

Customer churn is a situation that receives extensive analysis using a variety of techniques from data mining or machine learning. Data mining techniques may be used to anticipate customer churn. A data mining algorithm was used in this study to forecast client turnover. Logistic Regression and gradient boosting models were employed as one of the data mining techniques for implementing customer churn forecasts. The gradient boosting model could be said to perform better in predicting customer churn when compared to the logistic regression model, which produced accuracy training 87% and testing 88%. The results showed that the gradient boosting model was able to carry out the training and testing process with a training accuracy of 93% and a testing accuracy of 91%.