Intelligent Paid Subscription Renewal Prediction System Using Data Mining Techniques

Abstract

According to Virgo Capital, Typically, good services businesses have renewal rates of more than 80%, while more sticky software renewal rates hit 90% or more. Paid subscription trading websites collect huge amounts of customer’s data which, unfortunately, are not “mined” to discover hidden information for effective decision making. Hidden patterns discovery and relationships often go unexploited. This situation can be solved by using advanced data mining techniques. This research has developed a prototype Intelligent Paid Subscription Renewal Prediction System (IPSRPS) using data mining techniques, namely, Decision Trees, Naïve Bayes, and Neural Network. Each technique has its unique strength in realizing the objectives of the defined mining goals, which is shown in results. IPSRPS can answer complex “what if” queries which traditional decision support systems cannot. Using customer profiles such as the number of deals, sealed and un-sealed deals, profile interactions and the total sold amount it can predict the likelihood of customers renewing their subscription or not. It enables significant knowledge, e.g. patterns, relationships between service factors related to customer satisfaction, to be established. IPSRPS is Webbased, user-friendly, scalable, reliable and expandable. It is implemented on the .NET platform.