A variable selection method based on KPCA and FNN for nonlinear system modeling

The kernel principal components analysis (KPCA) can be used to convert a set of nonlinear variables into a linearly separable factors and overcome difficulties encountered with the existing multicollinearity between the factors. However the nonlinear system modeling method does not reduce the number of original features. This paper presents a novel method based on KPCA and selection of false nearest neighbor method (FNN) for secondary variables selection. In the proposed approach, it is inspired by FNN that interpretation of primary variable would be estimated by calculating the variables´ map distance in the KPCA space to select secondary variables. The results show that the method is effective and suitable for variable selection by comparing with the fully parametric model form the production processing of hydrogen cyanide.