The Performance Analysis of Male Handball Players Using Tree-Based Machine Learning Models

Each field of sports sciences includes several factors and variables that challenge the prediction and analysis of outcomes, such as game results or players' performances.Artificial Intelligence (AI) technologies, particularly machine learning (ML) algorithms, have the ability to perform tasks and solve problems that are complicated for human beings.One of these complicated tasks is the analysis of players' game and training performances of specific athletic tasks due to their nonlinear relationship and the diversity of specific types.Therefore, using AI techniques, particularly machine learning (ML) algorithms, contributes to analyzing players' performances and determining the significant factors that affect them.In this study, we analyze the performances of 40 male handball players using five tree-based ML models and determine the most critical factors influencing the performances of the countermovement jump with the hands-free skills of the players among ten demographic characteristics and physiological measurements.Initially, the prediction capabilities of the machine learning models are examined, and three superior models, Decision Tree, Gradient Boosting, and Extreme Gradient Boosting, are used to analyze the factors.The results suggest that the 10-meter sprint is the most influential factor on the players' performance, commonly for all ML models, while the 20-meter sprint is the least influential factor.This is one of the first studies using AI and machine learning for male handball players, and the findings are encouraging for future studies.