A new approach for robust fuzzy principal component analysis

The principal component analysis is negatively affected in the presence of outliers in the data. Therefore, we need to select a robust method, if the data include outliers. In this study, a new approach to robust fuzzy principal component analysis is proposed. This new approach bring together the strong sides of both robust and fuzzy methods. The performance of this approach is checked over a set of artificial data sets and an actual data set. The findings obtained in this study indicate that this new approach provide better results than the classical and robust principal component analysis. In the first section of the study, preliminary information on the analysis of classical principal component analysis and analysis of robust principal components and the literature summary are included. In the second section, some basic concepts and methods in the study are briefly explained. In the third section, a new approach algorithm for analyzing robust fuzzy principal components analysis and the data sets used in the study are introduced. In the next section, the findings obtained by applying the methods discussed in the previous sections are evaluated over the real and artificial data sets. In the last section, the results and suggestions obtained from the study will be discussed.