A confirmatory factor analysis approach for addressing the errors-in-variables problem with colored output noise

Abstract

Over the years, errors-in-variables (EIV) system identification has attracted considerable research interest. Among the many proposed approaches for identifying EIV models is confirmatory factor analysis (CFA), here referred to as EIV-CFA. This study extends previous research by presenting a EIV-CFA modeling framework that allows for colored output noise. Considerable attention is paid to the theoretical aspects of the minimum distance (MD) estimator. The finite sample performance of the MD estimator is briefly evaluated using simulation. The results suggest that model parameters are well estimated.

A confirmatory factor analysis approach for addressing the errors-in-variables problem with colored output noise