Selection of Key Performance Variables Through Principal Component Analysis and Their Behavior in Consecutive Handball Matches

Abstract

This study aimed to analyze spatiotemporal and mechanical workload variables during consecutive handball matches using unstructured data mining. Twentyeight national-level handball players participated in matches on two consecutive days, monitored using an ultra-wideband tracking system. Acceleration, deceleration, and high-speed variables were grouped using principal component analysis (PCA). Results revealed five different principal components for each match, except for the second match, which yielded six components. The PCA identified that variables related to acceleration, deceleration, and high-speed actions are crucial for understanding handball players' profiles, with the first principal component (PC1) explaining 49.2% of the variance. Differences were observed across most variables grouped in the PC1 between both competition days, finding higher values in the second day with effect sizes ranged from moderate to large (d = -0.31 to -0.79). This research provides valuable data on high-intensity mechanical and workload variables in handball, demonstrating how principal component analysis can be used to optimize performance in handball athletes. The findings offer practical implications for coaches and practitioners in designing training programs that consider these high-intensity physical demands characteristic of modern handball.