Multidimensional analyses to detect sport Talent Athletes with down syndrome

Abstract

In this paper we propose a methodology alternate to the somatotype to classify Down athletes in Peru, since they appear most alike in respect to this classification. A tandem analysis was performed, consisting in Principal Component Analysis of the ten body measures used for the somatotype, and Hierarchical Classification was run on the distances between athletes on the first two principal components through minimum variance clustering. A K-means partitioning was run further to optimize the classification. Three groups of measures resulted and six classes of athletes, whose differences could be checked via statistics. The results compared to the somatotype revealed more useful to identify a suitable training for the athletes to improve both their health conditions and sport performance.