Predicting Churn Customer in Telecom using Peergrading Regression Learning Technique
M.HemaLatha1, S.Mahalakshmi2
1Dr. M. Hema Latha*, Professor, Ramakrishna College of Arts & Science Coimbatore.
2S.Mahalakshmi, Research Scholar, Bharathiar University, Coimbatore.
Manuscript received on March 15, 2020. | Revised Manuscript received on March 31, 2020. | Manuscript published on April 10, 2020. | PP: 1025-1037 | Volume-9 Issue-6, April 2020. | Retrieval Number: F3861049620/2020©BEIESP | DOI: 10.35940/ijitee.F3861.049620
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Customer churn is an important issue which is faced by telecom industries daily. It is an essential concern for enterprises. Most of the telecom companies suffer a lot for voluntary churn. Here the churn rate is a significant impact for the industries on the lifetime value, and it affects the length of the service and also the revenue of the company. Because of the direct effect of the company revenue, especially in the telecom field, the companies are requesting to implement the development of predicting the potential customer to churn. Here the churn is an essential factor. Hence, analyzing the churn factor of increasing customer churn is as necessary as it is needed to reduce the churn. Recently telecom service facing the churn problem, for this analysis this research article focused on prediction of customer churn. In this research work, the main contribution is to develop a prediction model of churn, that supports the telecom operators to predict the customer, who is going to be churn. Here the model used is the Machine learning technique with the big data analysis. Our methodology used here is Peer Grading Regression (PGR). To provide a prediction of churn customer in telecom industries. To validate the performance of this proposed model, the Area under Curve (AUC) is used, and it is one of the standardized approaches. This model will be tested under some benchmark dataset; working on these vast datasets creates a transformation of raw data by telecom companies. The dataset values are provided for testing and training in the ensemble classifier. For experimentation, four algorithms are considered here: Peer Grading Regression (PGR), Random Forest, Decision Tree, and k-NN classifier. Moreover, the best outcomes are attained with the use of the boosting algorithm. Here this algorithm is applied for the classification of the churn prediction model.
Keywords: Machine Learning, Telecommunication, Customer Churn, Area Under Curve and Prediction
Scope of the Article: Machine Design