Robust Principal Component Analysis with Modified One-Step M-Estimator Method

Abstract

Abstract Principal component analysis (PCA) is a multiple variable analysis method that aims to reduce the dimensions of the original variable, which are mostly correlated so that new variables that are not correlated are obtained. The data used is criminality data in Indonesia in 2016 which contains outlier data in it. Therefore this study cannot use classic PCA because classic PCA was formed based on a covariant variant matrix that is very sensitive to the existence of outlier data. To overcome this problem, PCA robust is used with the Modified One-Step M-Estimator method with a MADn scale estimator to get the main components that are not much influenced by outliers. Modified One-Step M-Estimator (MOM) is the average remaining value of all extreme values that have been issued. The results obtained are there are 3 main components that can explain 85.19% of the variance of the 7 original variables.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint