Hybrid Conjoint Analysis - Symbolic Decision Tree Model For Customer Churn Prediction Model

The mobile phone market in Poland is a fast growing sector and according to forecasts in 2020 the whole telecommunication sector will be worth 47.78 bin PLN. The number of active sim cards has reached 53.3 min at the end of 2017. What is more, post-paid services are gaining more and more users and are more popular than pre-paid ones. Polish customers can easily change their mobile phone provider to another one. So the important issue arises - what factors can cause make customers to change their operator? In other words what causes customer churn in Polish mobile phone market? In order to take into account different factors (mainly preferences and other behavioral-like factors) a hybrid model will be used. This model uses a conjoint analysis based on stated preferences and combines it's results with a symbolic decision tree (where behavioral-like factors are used). Such approach allows to take into account different aspects of customer churn in one model - instead of many separated models - and allows to analyze all factors. Results allowed to detect most important customer churn factors.