Approach artificial neural network for nonlinear principal components analysis

Nonlinear Principal Component Analysis (NLPCA) is one of the explanatory dimension reducing technique and presents numerical and graphical results for variable set included linear or nonlinear relationships. Nonlinear Principal Component Analysis, categorical and ordinal variables as well as numerical variables can be included to analysis. Linearity assumption for observed variables does not need for Nonlinear Principal Component Analysis. In this study, artificial neural network approach for NLPCA was explained and applied. The hypothyroid data with 19 variables from 422 patients were used in the application. The results obtained using Principle Component Analysis (PCA) together with NLPCA were interpreted by presenting in tables and graphics. The first two principle components explained 95.65% of the total variance in the NLPCA, while they explained 90.08% of the total variance in the PCA. As a result, it was emphasized that NLPCA gives satisfactory results with a high variance explanation rate compared to PCA and thus it can be used for future predictions. Key Words: Hypothyroid, Nonlinear Principal Components Analysis, Principal Components Analysis