Chapter 8: Predicting Churn for Mobile Phone Providers

In Book 12: Operations Research Applications for Decision Analysis and Prediction

ISBN: 978-620-5-49772-2

Author: Vojo Bubevski (Independent Researcher)

Abstract 

This chapter presents a prediction of the churn for mobile phone providers. The term “churn” is a marketing term, meaning that a customer transfers loyalty from one provider to another. This prediction is particularly relevant in the industry, where companies try to keep their customers from churning. The prediction will give the company a better understanding of what behaviour leads to churning, and implement actions for preventing such behaviour. The snapshots of 1005 customer data are available, about 14% of whom have churned. Thus, this data is used to predict the Churn “Yes-No” variable. The Churn values for the first 15 customers are predicted by using Neural Network.

Keywords: Prediction, Risk Analysis, Mobile Phone Providers, Churn Prediction, Neural Networks.

 The Results

The prediction is that 10 customers will not churn (i.e., Churn is “No”) and five customers will churn (i.e., Churn is “Yes”). For example:

1.      Customer 1 will not churn with a probability of 0.941 (94.1%); There is a 0.059 (i.e., 5.9%) probability that this customer will churn;

2.      Customer 2 will not churn with a probability of 0.24 (24%); There is a 0.76 (76%) probability that this customer will churn; etc.

Prediction Model results are presented in Table 2.

Table 2: Prediction Results for 15 Customers