Regressing and predicting of multidimensional gait data characteristics——a novel correlation analysis method

In order to extract decision variables, conduct dimensionality reduction, make variable combination, and improve the regression of gait attribute model across complex data obtained from JiBuEn Gait Analysis System, ROF(Regression-Optimization-Feedback) is put forward in this paper as a novel correlation analysis method of multidimensional gait characteristics. LS-SVM(Least Squares Support Vector Machine) and FGO(Flexible Grid Optimization Algorithm) are combined to construct gait attribute model in the method. Meanwhile, a simplified experimental verification is proposed to discover strong relevant dimensions, and a better regression-prediction mode with the optimal combination of decision variables is presented in this paper. The experimental results show that ROF has a good character analysis effect as well as better value of practical application.