Abstract
In many real world applications data sets with missing values are quite common. In this paper, we propose a new approach which extracts local principal components for the feature extraction from a large scale database. The new method is based on a simultaneous approach to principal component analysis and fuzzy clustering with an incomplete data set including missing values. The simultaneous approach extracts local principal components by using the eigenvectors of the correlation coefficient matrix, while Fuzzy c-Varieties (FCV) proposed by Bezdek et al. uses the eigenvectors of the fuzzy scatter matrix.
