Predicting dropout and survival analysis in physical activity

Sports Science School of Rio Maior – Polytechnic Institute of Santarém
Life Quality Research Centre

Abstract

Physical activity is essential for people to keep a healthy lifestyle, in addition to the dimension of physical activity we also have the problem of sustainability and profitability of organizations that work in the context of sport. To contribute to the sustainability of organizations, we can approach customer dropout with an analytical perspective to support the development of measures to increase customer retention.

The use of historical customer data allows knowledge extraction through techniques based on machine learning, which enables the construction of predictive models. In the construction of predictive models, we can develop an analysis with a static perspective where we predict dropout at a given moment in time, or with a dynamic perspective, where we consider that the risk of dropout varies over time. We present an example, exploring a dynamic perspective and how can be used in sport management.

Survival analysis allows you to find moments where dropouts occur, and the use of the indicators obtained support decision-making for the development of new approaches to counteract dropout and contribute to a greater sustainability of sports organizations and sports practice.

Keywords: Retention; Machine learning; Decision trees; Survival analysis; Sport management.