Abstract

The driving motive behind this study is to provide the telecom industries with a novel tool for analysing their prominent data. An attempt has been made to develop a classification model, which uses the telecom operator's data to predict the customer's behaviour. Every day, there are a huge number of people churning away from their telecom service provider. This work is centred around finding those customers who will churn, and their reason for churning. Several classification models were considered. After comparing all the models, Support Vector Machine and an ensemble model of Random Forest were finalized as the classifiers to be used for churn classification. This decision was made based on the F1-score. Applying such methodologies, resulted in a more efficient, fully functional classifier that can predict customer churn. The list of churning customers is given to the telecom provider along with the most likely reason for their churning. The reason for the churning of the customer has been found using K-means clustering and analysing various graphs such as, histograms, bar graphs and boxplots. Using this information, the telecom operator can come up with feasible plans to stop the customer from churning. The classification model developed in this study can strengthen the telecom provider. Such a novel and automated model is essential for every telecom giant in the long run.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call