Churn prediction for gym members using artificial neural networks assisted with the psychological concept of habit formation in the fitness industry

Abstract

Because of the growth of the health and fitness awareness, the gym industry is booming. Consequently, the competition between gym businesses is becoming increasingly fierce. So, it is vital for gyms to develop membership retention strategies to keep their customers from switching to a competitor. An important step in such strategies is to proactively detect members who are vulnerable to ending their memberships (Churning) to make efforts into retaining them. Utilizing technology for predicting churn can be advantageous for fitness clubs that aim to stay at the top of the business. Hence, the purpose of this paper is to develop a model that predicts gym membership churn by employing multi-layer perceptron (MLP) artificial neural networks (ANN) using backpropagation with an emphasis on feature selection. Two methods were used to base feature selection on: Literature review and filtering method. The results found that implementing the psychological concept of habit formation served as a link for introducing an effective ANN model into the fitness retention strategies, as the model achieved high prediction performance metrics, namely, accuracy, sensitivity, and specificity, 92.1%, 89.1% and 93.8% respectively.