Data-Driven Customer Retention Strategies in E-commerce: A Fuzzy Z-Number Approach

Customer attrition continues to be an issue in the e-business environment, which may result in considerable losses in terms of sales. This research proposes a new hybrid model for predicting and controlling customer churn in a more efficient way. Our model, which uses both machine learning and fuzzy logic techniques, provides a stable solution to this important issue. For the purpose of evaluating customer churn, we use the Support Vector Machine (SVM) model. In order to improve the performance of the SVM, we tune its parameters using the Coyote Optimization Algorithm (COA). After customer churn has been predicted, Fuzzy Z-numbers are used to help make the right decisions on which customers should be retained. This approach enables us to bring in quantitative and qualitative aspects of the decision-making process like the customers’ behavior, their preferences, and demographic data. The performance of the proposed model is thoroughly tested on a dataset of 8,913 transactions from the WooCommerce platform. The outcomes show better efficiency in terms of precision, which is equal to 0.98, and accuracy of 97.2%, showing improvements compared to the previous methods. This research offers a useful instrument that will help e-commerce companies improve their customer retention strategies and achieve sustainable growth.