Churn prediction in the telecom business

Abstract

Telecommunication companies are acknowledging the existing connection between customer satisfaction and company revenues. Customer churn in telecom refers to a customer that ceases his relationship with a company. Besides losing the customer, it is also likely that the customer will join a competitor company. These companies rely on three main strategies to generate more revenue: acquire new customers, upsell the existing ones or increase customer retention. Some articles suggest that churn prediction in telecom has recently gained substantial interest of stakeholders, who noticed that retaining a customer is substantially cheaper that gaining a new one. Customer churn gets special attention by telecom service providers. The easiness in changing operators and Mobile Number Portability allow customers to switch to another provider easily and reserving their telephone numbers. Many approaches were already developed for predicting churn rates in telecom business. This dissertation compares multiple approaches and algorithms that identify identify the clients who are closer to abandon their telecom provider. For the purpose of this research, real data was be provided by WeDo technologies, which is the number one provider of revenue assurance and fraud management software for telecom operators. The use of real data extended the refinement time necessary, but ensured that the developed algorithm and model can be applied to real world situations. The large dataset available opens a new set of possibilities, and makes it possible to obtain interesting and novel results.