Hybrid machine learning techniques for predicting customer churn

Abstract

With global advancement, Information Technology has led to the growth of numerous Service Providers, which, in turn, has resulted in fierce competition between themselves. For Service Providers, the most prevalent obstacle is the handling of customer churn, retention, and satisfaction of customers for successful market sustenance. Customer Relationship Management (CRM) concentrates on boosting, sustaining, and building long term customer associations. CRM relies on the collection of information prior to making decisions. When a customer halts the existing service provider relationship and shifts to another, this is referred to as churn. The overall business profit and image are perturbed by this never-ending motion of churning. Therefore, it is more preferable to stop customers from churning and going for churn prediction. The main challenge of churn prediction is the absence of a single cause for customer churn. This is generally an accumulation of various other causes. It is very complicated to detect these causes, as they are reliant on the organizational services used by the customers and also on the individual views of the customer. In this area of expertise, the key necessity of organizations is early churn prediction, identification of the main reasons for churn, and the prediction of countermeasures to prevent churn. The available organizational data can be employed for all these actions. However, the prediction mechanism is greatly hindered by the data s nature. Tangible or intangible service-oriented areas, product-based businesses, and telecommunication services are churn prediction application areas. onventional churn prediction techniques have the advantage of being simple and robust with respect to defects in the input data, they possess serious limitations to the interpretation of reasons for churn. Therefore, measuring the effectiveness of a prediction model depends also on how well the results can be interpreted for inferring the possible reasons for churn. The churn models aim to identify the early churn signals and recognize the customers with a high likelihood of voluntarily leaving. In this work, a methodology for evaluating statistical models for classification with the use of a composite indicator is proposed. With extensive research in Artificial Intelligence, it has become possible to dig to the core of the factor responsible for customer churn. newline