An Rn Surface to Describe Sport Performance in Women's Basketball Using Kohonen Nets on ECG Signals

Abstract

In this paper, we describe a procedure to create a surface based on R29 information, each vector corresponds to information taken from players of women basketball team of Universidad Militar Nueva Granada, those data include frequency data from ECG, statistic data from heart rate variability, and anatomic information as age and height. A close following was done for each player conforming the team, a total of 96 registers from every single player were acquired and used for create a 29 vector to characterize each player in a specific date. Konohen net technique was employed to make a mapping between high dimensional data and 2 dimensional system of coordinates and nodes, after that, information resulting from Kohonen net was used to represent the performance on the complete team, and groups were detected into the map that can describe and even predict some performance characteristics of a new player.