Customer Relationship Management (CRM) is a strategic approach which targets the development of profitable, long-term relationships with key customers and stakeholders. A major concern in CRM in telecommunications companies is the ease with which customers can move to a competitor, a process called “churningâ€. Churning is a costly process for the company, as it is much cheaper to retain a customer than to acquire a new one. Customer churn has emerged as one of the major issues in Telecom Industry. Telecommunication research indicates that it is more expensive to gain a new customer than to retain an existing one. This research works a real-world study on customer churn prediction and proposes the use of classifier to enhance a customer churn prediction model. The major objective of this application was to predict customer churn behaviour of individual customers. A critical success factor for this proposed work was clever preprocessing of the given data, in particular the construction of derived predictor features. The first step is to understand the data that serves commercial values. Data preparation entails preprocessing of the raw data to converts incomplete dataset into complete dataset by using Ascent Monte Carlo Expectation Maximization (AMCEM) Algorithm which containing limited information. In the customer churn prediction analysis, sometimes heavy customers also tend to churn. In order to analyze partial and total defection, this study defines changes in a customer’s status from active use to non-use or suspended as partial defection and from active use to churn as total defection. Thus, mediating effects of a customer’s partial defection on the relationship between the churn determinants and total defection are analyzed by using the Fuzzy Poisson Naive Bayes (FPNB) model. Then customer churn behavior analysis is applied to FPNB model with four conditions like customer dissatisfaction (), switching costs (), service usage () and customer status (). The attributes are derived from call details and customer profiles, those attributes are selected using Hybrid Cuckoo Harmony Search (HCHS) algorithm. Multi-Kernel Extreme Learning Machine (MKELM) classifier is used in this work as a basis prediction model for customer churn prediction. The experimentation results of the proposed MKELM prediction model is compared to single logistic regression model, boosting algorithm, Support Vector Machine (SVM) classifier and Kernel Extreme Learning Machine (KELM) classifier. Experimental results demonstrated that PFNN also provides a good separation of churn data, highly scalable, so, MKELM is used for churn prediction examination in the telecommunication industry.
Volume 9 | 12-Special Issue
Pages: 567-582