Customer Churn Prediction In Telecommunication Industry Using Random Forest Classifier

Abstract

Nowadays data has become the important aspect in each and every field. In this the data about the telecommunication industry is collected and then the raw data is classified into churn and the non churn customers. The churn customers are one who periodically uses the same resource signals and non churn customers are one who utilizes the resources based on the services provided by the particular company. In existing system they uses the algorithm called LDT and UDT which train the system blindly with too many attributes which are not necessary for the computation. So it takes much time to train the system and the accuracy is not that much efficient and it achieve the performance about 84 percent. But this much of performance is not that much efficient for an organization to provide convincible services. So in order to resolve this problem in existing system we proposing the system with an efficient algorithms known as Random Forest Classifier and Support Vector Machine which selects the important attribute which increases the performance of the system and by implementing these two algorithms we can achieve the efficiency of about 95 percent. Because this efficiency in performance will ensure the company to provide the appropriate services to retain the non churn customer within the organization to sustain the Telecommunication industry.