A holistic approach to performance prediction in collegiate athletics: player, team, and conference perspectives
Christopher B. Taber,Srishti Sharma,Mehul S. Raval,Samah Senbel,Allison Keefe,Jui Shah,Emma Patterson,Julie Nolan,N. Sertac Artan,Tolga Kaya
DOI: https://doi.org/10.1038/s41598-024-51658-8
IF: 4.6
2024-01-13
Scientific Reports
Abstract:Predictive sports data analytics can be revolutionary for sports performance. Existing literature discusses players' or teams' performance, independently or in tandem. Using Machine Learning (ML), this paper aims to holistically evaluate player-, team-, and conference (season)-level performances in Division-1 Women's basketball. The players were monitored and tested through a full competitive year. The performance was quantified at the player level using the reactive strength index modified (RSImod), at the team level by the game score (GS) metric, and finally at the conference level through Player Efficiency Rating (PER). The data includes parameters from training, subjective stress, sleep, and recovery (WHOOP straps), in-game statistics (Polar monitors), and countermovement jumps. We used data balancing techniques and an Extreme Gradient Boosting (XGB) classifier to predict RSI and GS with greater than 90% accuracy and a 0.9 F1 score. The XGB regressor predicted PER with an MSE of 0.026 and an R 2 of 0.680. Ensemble of Random Forest, XGB, and correlation finds feature importance at all levels. We used Partial Dependence Plots to understand the impact of each feature on the target variable. Quantifying and predicting performance at all levels will allow coaches to monitor athlete readiness and help improve training.
multidisciplinary sciences