Customer Profiling and Retention Using Recommendation System and Factor Identification to Predict Customer Churn in Telecom Industry

Abstract

In the telecommunication area, an immense bulk of data is being created consistently because of a huge customer base. Chiefs and business analyst stressed that accomplishing new customer is prohibitive rather than retaining the current ones. Business experts and CRM analyzers need to know the explanations behind customer's attrition, just as, behavior patterns from the current churn customers so that the business applies some retention strategies to retain the customers before they leave the industry. This paper proposes a, customer churn prediction model that utilizes numerous machine learning classification algorithms, like gradient boost, random forest (RF), decision tree (DT), logistic regression (LR), K-neighbors, in which the gradient boost classifier outperformed other classifiers, and, clustering technique has been utilized to break down and recognize the churn customers and give the factors behind the churn customers in the telecommunication sector. The proposed model initially classifies churn customers data, utilizing classification algorithms, After classification, the proposed model sections the churn customers data by categorizing the customers in groups utilizing k-means clustering algorithm to give group-based retention offers. This paper additionally distinguished churn factors that are basic in deciding the underlying drivers of churn in which the “attribute selected classifier” algorithm has been utilized from Weka tool; by realizing the critical churn factors from customer’s data, CRM can improve profitability, prescribe pertinent advancements and relevant promotions to the group of likely churn customers, dependent on similar behavior patterns, and excessively improve marketing campaigns of the organization (Ullah, Irfan, et al. “A churn prediction model using random forest: analysis of machine learning techniques for churn prediction and factor identification in telecom sector.” EEE Access 7 (2019): 60, 134–60, 149). In addition, recommendation system has been utilized for customer retention. As creating retention strategies is a basic undertaking of the CRM to prevent customers from attrition. The proposed churn prediction model is assessed utilizing measurement matrices, for example, accuracy, precision, recall, f-measure, and receiving operating characteristics (ROC) area. The outcomes uncover that the proposed customer churn prediction model delivered better churn classification utilizing the gradient boost algorithm and customer profiling, utilizing k-means clustering. Moreover, it additionally gives factors behind customer’s turnover, through the rules produced by utilizing the attribute selected classifier algorithm. In the end of the examination, some retention strategy has been applied using recommendation system to maintain and keep up the customers for long time base.