Applying client churn prediction modeling on home-based care services industry

Client churn prediction model is widely acknowledged as an effective way of realizing customer life-time value especially in service-oriented industries and under a competitive business environment. Churn model allows targeting of clients for retention campaigns and is a critical component of customer relationship management(CRM) and business intelligence systems. There are numerous statistical models and techniques applied successfully on data mining projects for various industries. While there is literature for prediction modeling on hospital health care services, non-exist for home-based care services. In this study, logistic regression, random forest and C5.0 decision tree were the models used in building a binary client churn classifier for a home-based care services company based in Australia. All models yielded prediction accuracies over 90% with tree based classifiers marginally higher and C5.0 model found to be suitable for use in this industry. This study also showed that existing client satisfaction measures currently in use by the company does not adequately contribute to churn analysis.