Multi-Output Regression for the Prediction of World-Class Performances in Women’s Handball

The purpose of this study was to investigate the relationship between workload and in-game technical and athletic performance. To achieve this,A modeling approach that predicts multiple numerical output variables simultaneously, particularly useful when these outputs are correlated (multi-output regression) models were used to predict 7 performance indicators based on previous training and game athletic workloads measured by Inertial Measurement Units (IMU) indicators, previous in-game actions annotated by staff members and game contextual factors. We compared 4 single-output models (kNN, regression tree, random forest and(NN) Predictive models inspired by the human brain, used in this study for multi-output prediction in sports performance analysis (neural networks)), their multi-output counterparts and aA baseline model predicting future performance as the average of each player’s past performance, serving as a simple reference for comparison with more complex models (dummy baseline) (predicting the average performance of each player over the last month) in terms of average(Root Mean Squared Error) A measure of the quadratic difference between predicted and actual values in regression models (RMSE) (aRMSE) during aAn evaluation method where past training and game data are used sequentially to predict performance of the next game (chronological evaluation) where previous trainings and games data are used to train models to predict the next game performances. Overall, the use of multi-output regression models enabled a decrease of the average predictive error (A metric for prediction error that evaluates model accuracy in terms of average squared errors across multiple outputs in a multi-output models (aRMSE) = 4.23) in regards to their single-output counterparts (aRMSE = 4.35) while providing a significant decrease of average computation times (4.75 to 0.82 seconds). Among the 4 multi-output models, only the kNN (aRMSE = 3.852) and random forest (aRMSE = 3.888) performed better than the dummy regressor (aRMSE = 3.944). These results point towards that physical training may have a limited impact on game performance.