A Data-Driven Approach to Customer Churn Prediction in E-Commerce Business Using Machine Learning

Customer churn is a crucial factor which is affecting businesses in multiple industries. Even if one customer leaves at a time, it's still a significant loss for business. This study focuses on the uses of different machine learning algorithm to find out customer churn in e-commerce business. As business in online platform is growing rapidly, customer churn has become a critical challenge which leads businesses to find effective ways for customer retention. In this research, different models of ML such as the Random Forest, XGBoost, Logistic Regression, Gradient Boosting, SVM, LightGBM, and KNN used to evaluate customer churn prediction. This study also applied the SMOTEENN technique for handling the imbalance of class and implement feature selection approaches for improving the model efficiency. The results show the effectiveness of LightGBM which achieve an accuracy of 98.265% and perform better than other models in the matter of precision, accuracy, F1-score, and recall. Results from this research, give valuable insights for business to develop different retention strategies, improve customer feedback, and optimize profitability.