APPLYING MACHINE LEARNING TO ANALYZE ATHLETE RETENTION IN SPORTS PROGRAMS: A CUSTOMER CHURN STUDY

Abstract

Athlete retention is a critical challenge within sports programs, where understanding the factors leading to athlete churn can significantly enhance management strategies and improve program offerings. Traditional methods of predicting athlete churn have struggled due to the complexity and vast scale of data involved. Machine learning presents a more robust approach, offering the ability to handle high-dimensional data and complex relationships more effectively than simple linear models. This study explores the application of advanced machine learning techniques to athlete churn within the sports sector, benchmarking 16 state-of-the-art models in handling data imbalance, a common issue in churn prediction. Contrary to traditional reliance on the Synthetic Minority Oversampling Technique (SMOTE) to address data imbalances, our findings suggest that SMOTE may not be universally effective. In cases where certain models already perform exceptionally, SMOTE's application might lead to overlooking nuanced data segments or fail to differentiate between overlapping data points. This analysis provides a comprehensive review of machine learning strategies to optimize athlete retention strategies, offering insights into the practical limitations and potential of various models in the sports industry context.