11 International Journal of Advances in Computer Science & Its Applications Volume 6 : Issue 1 [ISSN 2250-3765] Publication Date : 18 April, 2016 Churn Prediction Retention Framework Essam Shaaban Aboul Ella Hassanien Abstract- Churn prediction is considered a big issue in the Telecom market because customer acquisition costs five to eight times than retaining an existing one. Customer retention is the action that a service provider undertakes in order to reduce customer dissatisfaction and decrease the probability of leaving a service provider. This paper proposes a churn prediction retention framework based on data mining techniques. Deploying the proposed framework into a business intelligence system can help in enhancing the efficiency of customer relationship management. Moreover it can help customer churn management department to easily predict and retain the expected future churners in many business areas. The proposed framework mainly composed of seven phases and sub-phases. A case study is demonstrated to evaluate the framework throughout 5000 customers’ records in an anonymous telecom company. Keywords- Churn prediction, customer retention, data mining, clustering, classification I. Introduction The purpose of prediction is to anticipate the value that a random variable will assume in the future or to estimate the likelihood of future events [1]. Churn prediction is a useful tool to predict customer at churn risk. Customer Churn Management (CCM) becomes one of the most critical success factors mainly due to higher acquisition costs for new customers [2]. The term churn management has been adopted to define customer turnover [3]. Besides; CCM is the concept of identifying those customers who are intending to move their custom to a competing service provider. Retaining an existing customer from being a churner is the best core strategy to survive in industry as well as it is becoming a common knowledge in business [4], [3]. The prevention of customer churn through customer retention is a core issue of Customer Relationship Management (CRM). Customer retention is defined as the practice of working to satisfy customers with the intention of developing long term relationships with them [5]. Retention is a result of customer desertion. Urquizo [6] stated that the ways in which customers leave are; credit does not sufficiently help customers achieve their goals, Essam Shaaban Faculty of computers and information, Beni-Suef University Egypt Aboul Ella Hassanien Faculty of computers and information, Cairo University Egypt customers are being negatively affected and are having difficulties making payments and customers find alternative service providers. Regarding this research, customer retention can be defined as the organizational plans, actions and activities to retain potential customers by developing, maintaining and maximizing mutually beneficial long-term relationships [7]. It is supported by many factors, including: high customer churn rate, evolutionary needs and behavior of customers, increasing national and international competition, mature markets, and inappropriateness of traditional retention analysis tools to tackle the problem of having a churn prediction retention model in Business Intelligence (BI) applications that can efficiently identify churners in telcos and provide a retention solution. The aims of this paper are to analyze the existing prediction and retention models to highlight their capabilities and limitations. Then develop a framework for Pre-paid mobile churn prediction and retention. This paper is organized as follows. Section II describes the theoretical background in charge of prediction models, customer retention, and limitations of churn prediction and retention models. Section III shows in a detailed format the proposed churn prediction retention framework. Section IV provides the results of an implemented case study. Finally; conclusion and future work are presented. II. Theoretical Background A. Prediction Models A predictive model is defined as one that takes patterns that have been discovered in the database, and predicts the future [8]. Predictive modeling is mainly concerned with predicting how the customer will behave in the future by analyzing their past behavior. As Figure 1 depicts, prediction models can be classified into two types; traditional models and soft computing models. Figure 1: Prediction models The most popular traditional prediction models are; Decision Tree and Regression Analysis. Decision tree is the most popular type of predictive model. It has