DATA REDUCTION AS A NECESSARY TOOL IN BIOMECHANICAL KINEMATIC DATA ANALYSIS

Abstract

The purpose of this study was to evaluate four possible methods of data reduction for the subsequent use of the data with an Artificial Neural Network (ANN). Functional data analysis and dynamical systems theory approaches were used to investigate the kinematic dynamics of gait and rowing movements. Five rowing participants completed a 2000m row on a RowPerfect ergometer and 24 gait participants completed treadmill running. The results indicate that each of the four methods provides possibilities for use in an ANN by utilising data reduction. In particular the continuous relative phase, with its use of two joint position and velocity, can compress four variables into one, and can maintain the trends of the data.