Analyzing game statistics and career trajectories of female elite junior tennis players: A machine learning approach

Tennis is a popular and complex sport influenced by various factors. Early training increases the risk of career dropout before peak performance. This study analyzed game statistics of World Junior Tennis Final participants (2012–2016), their career paths and it examined how game statistics impact rankings of top 300 female players, aiming to develop an accurate model using percentage-based variables. Descriptive and inferential statistics, including neural networks, were employed. Four machine learning models with categorical predictors and one response were created. Seven models with up to 18 variables and one ordinal (WTA rank) were also developed. Tournament rankings could be predicted using categorical data, but not subsequent professional rankings. Although effects on rankings among top 300 female players were identified, a reliable predictive model using only percentage-based data was not achieved. AI models provided insights into rankings and performance indicators, revealing a lower dropout rate than reported. Participation in elite junior tournaments is crucial for career development and designing training plans in tennis. Further research should explore game statistics, dropout rates, additional variables, and fine-tuning of AI models to improve predictions and understanding of the sport.