PRINCIPAL COMPONENT ANALYSIS OF KNEE ANGLE WAVEFORMS DURING RACE WALKING

Abstract

This study aimed at understanding whether principal component analysis (PCA) may be useful to characterize race-walkers abilities at different performance levels. Seven young race-walkers of national and international rank were recruited. PCA was applied for classifying and detecting the structure of knee sagittal angle. This statistical technique allowed extracting multidimensional features that capture the greatest variation in race walking data. The scores, i.e. the projections of the original data on the components, revealed to be good discriminative factors for performance level detection. Finally, the underlying linear structure of the principal components provided a biomechanical interpretation of motor skill. The best athletes were able to correctly lock the knee during the mid-stance; the worst ones tended to bend the knee prematurely.