فهرست نشریات علمی دسترسی آزاد ایران

Given the extensive changes in today’s business landscape, the continuous development of technology, and the need to pay attention to customers and respond to their demands in order to preserve organizational survival, preventing customer churn is necessary and unavoidable. Therefore, retaining existing customers, attracting new customers, and controlling customer churn for business growth and development is essential. The aim of this research is to present an algorithm to prevent customer churn in e-commerce based on feature selection using metaheuristic methods and deep learning. This research, in terms of its objective, results, and algorithm development, is based on operational research methods. The research variables are composite and the study is cross-sectional in time. The results indicate that the proposed algorithm is capable of finding the best approach for predicting customer churn. In this research, first, the features of customer churn are selected using genetic optimization algorithm based on binary human learning, and then, selected features as input of a deep learning model are used to predict customer churn in e-commerce. The research findings indicate that the optimization and deep learning algorithm, with an accuracy of 92.32, performed better than other customer churn prediction methods and can, as an efficient algorithm, assist organizations in improving performance.