Anthropometric and Motor Fitness Based Assessment of Playing Positions in Volleyball Players with the AID of Predictive Machine Learning Models

Abstract

Volleyball is a team sport in which the performance of the players is often dependant on various factors such as regular training and playing positions which are in turn affected by several factors of players. The Anthropometric Parameters (AP) indicate the body composition of the individual and can be used to ascertain the suitable playing positions of players. Further, aspects such as Motor Fitness Parameters (MFP) can impact the quality of play in volleyball. The present work was successful in concluding that the BMI and Height in AP and Explosive Power (EP) and Relative Jump (RJ) in MFP are indicative of playing positions, with EP and RJ being statistically significant features as well. For predicting suitable playing positions, machine learning algorithms namely Support Vector Machine (SVM), SVM with variable scaling, SVM with hyper parameter optimization and Extreme Gradient Boosting (XG Boost) with model based learning parameters were used. The classification results were found to be accurate upto 98.98% in SVM with tuned hyper parameter optimization technique and in XG Boost. But XG Boost was found to perform significantly faster than the former approach. Such approaches can be incorporated in various training and rehabilitation programs in volleyball to improve the performance of the players.