Predictive churn analysis with machine learning methods

In this study, we analyzed the well known machine learning algorithms which are mostly used in the past studies to design a new model to predict customer churn. In addition to logistic regression, Naive Bayes classifier, desicion trees, support vector machines and artificial neural Networks, we designed a new hybrid model and analze the initial results. We analyzed that the success of the hybrid method developed by using logistic regression and Naive Bayes methods is higher than the success rates that is obtained when these algorithms are applied alone. On the other hand, artificial neural networks method showed the highest prediction rate %91 as expected.