Archives

A Churn Detection Model in Telecommunication Using Machine Learning Techniques


Ch. Ratna Jyothi, Neha Basavana boyina, S. Vijaya Lakshmi, B. Akhila, A. Rahul Sai
Abstract

As technology increases in recent days, business became highly increased day by day especially in the field of telecom domain, a huge volume of data is being generated daily due to a heavy usage from the customers. Many Telephone service companies often use customer churn or moving analysis and customer churn rates are one of their metrics in business churn detection because the cost of getting new customer is costlier than maintaining the existing ones. Decision analysers and executives need to know the explanations behind turnover or migration of the clients. This paper proposes customer model for churn customers to another service that utilizes some classification algorithms to see the moving of clients and gives the most affecting elements behind the turnover of clients in the telecom segment. Highlighting the most impacting features is performed by utilizing Recursive Feature Elimination(RFE) and Variance Inflation Factor(VIF).The proposed model classifies the churn clients information utilizing supervised classification algorithms, in which the Support Vector Machine(SVM) with Synthetic Minority Oversampling Technique(SMOTE) and Principal Component Analysis(PCA) i.e SVM+PCA+SMOTE calculation performed well with 97.11% exactness than logistic regression, logistic regression +PCA+SMOTE, Random Forest +PCA+SMOTE . This paper additionally recognized the most affecting elements for churning that are basic utilized in deciding the underlying factors of turnover. By knowing the huge churn factors from the information, the profitability of the organization increments. The proposed churn recognition model is assessed by utilizing metrics such as accuracy, sensitivity(true positive rates) and specificity(true negative rates). The outcome for our proposed churn detection model produced better churn classification using the SVM+PCA+SMOTE .

Volume 12 | Issue 2

Pages: 936-943

DOI: 10.5373/JARDCS/V12I2/S20201118