Churn Prediction with Sequential Data Using Long Short Term Memory

Abstract

Companies aim to keep their current clients in today's dynamic market climate. Churn prediction is essential in order to accomplish that. It is not a straight forward task to predict churning consumers. In the fast-food industry, it is even more difficult since, when a consumer stops having the industry's products, there can be different reasons. The study is proposed to resolve this situation. The customer data structure is sequentially established with the data related to the customers. With the sequential data, a long short term memory model is designed to estimate the customers' churn stages and compared with the conventional methods of classification. The proposed model offers promising results and stands out among related studies with its predictions.