ANALISIS REGRESI KOMPONEN UTAMA ROBUST UNTUK DATA MENGANDUNG PENCILAN

Abstract

Principal Component Regression (PCR) is one of the widely used statistical techniques for regression analysis with
colinearity, and a robust technique on PCR when data contains outlier is an important problem. In this paper we consider
the problem of robust PCR based on Minimum Volume Ellipsoid (MVE) estimator and Least Trimmed Square (LTS)
regression. We aimed to look at the behavior of the principal component regression coefficient resulted by MVE-LTS and
compare them with classical estimator through the bias and the mean square error. The result shows that PCR using
MVE-LTS is very robust.
Keywords : Principal component regression, colinearity, robust