Extracting Actionable Knowledge to Increase Business Utility in Sport Services

Abstract

The increase in retention of customer in gyms and health clubs is nowadays a challenge that requires concrete and personalized actions. Traditional data mining studies focused essentially on predictive analytics, neglecting the business domain. This work presents an actionable knowledge discovery system which uses the following pipeline (data collection, predictive model, loyalty actions). In the first step, it extracts and transforms existing real data from databases of the sports facilities. In a second step, predictive models are applied to identify user profiles more susceptible to dropout. Actionable rules are generated based on actionable attributes that should be avoided, in order to increase retention. Finally, in the third step, based on the previous actionable knowledge, experimental planning is carried out, with test and control groups, in order to find the best loyalty actions for customer retention. This document presents a simulation and the measure of the business utility of an actions sequence to avoid dropout.