Analisis Customer Churn Menggunakan Bayesian Belief Network (Studi Kasus: Perusahaan Layanan Internet)

Abstract

The number of Internet users around the world tends to increase every year. However, the growth is not always accompanied by an increase in economy of scale for all types of telecommunication companies. For example, the Internet Service Providers have churn rates up to 30% peryear. Therefore, many research on customer churn at telecommunications industry have been conducted. Among of these studies, there has no research specifically performed to analyze customer churn using Bayesian Belief Network (BBN) for the ISP. In this study, customer churn analysis for the Internet service providers was conducted by involving BNN method. Yet, prior to applying BNN, an outlier analysis was employed to eliminate extreme outliers that may contain in the data. The overall stages of customer churn analysis used in this study include data preparation, continuous variable discretization, outlier analysis, correlation analysis, multicollinearity test, and construction of causality and BBN diagram