fbpx
×

Customer Churn Analysis and Prediction in Telecommunication for Decision Making

  • P.K.D.N.M. Alwis ,Sabaragamuwa University of Sri Lanka Belihuloya, Sri Lanka (madushani.niroshi@gmail.com)
  • B.T.G.S. Kumara ,Sabaragamuwa University of Sri Lanka Belihuloya, Sri Lanka
  • H.A.C.S. Hapuarachchi ,Sabaragamuwa University of Sri Lanka Belihuloya, Sri Lanka
Abstract

With the rapid development of communication technology, the field of telecommunication faces complex challenges due to the number of vibrant competitive service providers. Customer Churn is the major issue that faces by the Telecommunication industries in the world. Churn is the activity of customers leaving the company and discarding the services offered by it, due to the dissatisfaction with the services. The main areas of this research contend with the ability to identify potential churn customers, cluster customers with similar consumption behavior and mine the relevant patterns embedded in the collected data. The primary data collected from customers were used to create a predictive churn model that obtain customer churn rate of five telecommunication companies. For model building, classified the relevant variables with the use of the Pearson chi-square test, cluster analysis, and association rule mining. Using the Weka, the cluster results produced the involvement of customers, interest areas and reasons for the churn decision to enhance marketing and promotional activities. Using the Rapid miner, the association rule mining with the FPGrowth component was expressed rules to identify interestingness patterns and trends in the collected data have a huge influence on the revenues and growth of the telecommunication companies. Then, the C5.0 Decision tree algorithm tree, the Bayesian Network algorithm, the Logistic Regression algorithm, and the Neural Network algorithms were developed using the IBM SPSS Modeler 18. Finally, comparative evaluation is performed to discover the optimal model and test the model with accurate, consistent and reliable results.

Keywords—bayesian network, c5.0 decision tree, logistic regression, neural network

Please fill the following information to view the document.



TOP