HYBRID APPROACH OF BOOSTED TREE FOR CHURN PREDICTION IN MATLAB

Organization puts much attempt to hold the churn clients in the company by identifying them as clients are beneficial persons to the growth of a company. Hybrid approach of Boosted tree is one of advance algorithm in data mining for the prediction of churn clients. The Hybrid approach of Boosted tree involved J48 and LogitBoost with some parameters. This Hybrid approach uses the feature of PSO that helps to predict the churn clients with more accuracy. This paper enhances the boosted tree with hybrid approach and produces more accurate results in the prediction of churn clients.  The results obtained with Hybrid approach shows that the new approach is better in predicting the churn clients from existing boosted tree technique.