Abstract
Amidst the booming development of e-commerce and intense market competition, numerous e-commerce companies frequently encounter the issue of customer loss. This research endeavors to offer a comprehensive analysis and precise forecasting of customer churn behavior for an E-commerce company. The research utilizes the E-commerce Customer Churn dataset From Kaggle, which offers a wealth of customer information. The paper initially performs a data cleaning to fill the missing value by K-nearest neighbors (KNN). And then, it also performs feature engineering to preprocess the dataset. Subsequently, multiple machine learning models were constructed, including Logistical Regression (LR), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), Neural Network (NN), and a stacking model with a metal-leaner as Extreme Gradient Boosting (XGBoost) has been developed. The stacking model achieved the highest performance with 92.8% accuracy and 0.940 AUC. Key factors such as tenure, complaints, cashback amount, order recency, and satisfaction score were identified as important predictors. This research demonstrates the potential of Machine Learning in developing effective retention strategies for e-commerce platforms.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call