Customer churn prediction in telecom using big data analytics

Abstract

Abstract Customer churn will cause huge losses to the communication company and has become a real problem. The article uses big data analysis technology to analyse user characteristics of churn customer historical information data, establish a churn prediction model, find users with a higher risk of churn in advance, develop targeted strategies, and carry out a series of retention activities to retrieve them. The paper presents a strategy of user segmentation and piecewise regression to find the highly relevant fields and divide the customers into different groups based on these fields, and then use regression analysis to establish the prediction models for different groups. Online test shows that the model can effectively identify most of the lost customers, effectively reduce the user off-network rate, and improve efficiency and effectiveness than traditional methods.