The fastest algorithm for analyzing robust principal    components with application on variables affecting                    the increase of aluminum level in blood Abstract

Y.Khmo And Dhafer H.Rashid, Khlod

doi:10.32894/kujss.2006.44074

The fastest algorithm for analyzing robust principal components with application on variables affecting the increase of aluminum level in blood Abstract

Author

Khlod Y.Khmo and Dhafer H.Rashid

10.32894/kujss.2006.44074

Abstract

One of the common techniques in analyzing multivariate data is analyzing of the principal components. This transforms large number of related variables into lesser number of no related components (PCA).
In case of existence of outliers, which can be detected in several ways, then the dependence of variance and ordinary covariance matrices, also correlation matrix, would lead to misleading results in analyzing the principal components.
The aim of this research is to introduce a new and fast algorithm in analyzing the robust principal components; when data contain outliers, while conventional methods fall in detecting outliers in data; then the results are misleading. The method is implemented to show its real effectiveness on variables affecting the increase of aluminum level in blood.

Keywords