Errors-in-variables (EIV) identification refers to the problem of consistently estimating linear dynamic systems whose output and input variables are affected by additive noise. Several estimation methods have been proposed for identifying linear dynamic systems from noise-corrupted output measurements. This talk introduces Structural Equation Modeling (SEM) setting to EIV identification. Two schemes for how EIV Single-Input Single-Output (SISO) systems can be formulated as SEMs are presented. The proposed formulations allow for quick implementation using standard SEM software. Simulation examples show that compared to existing procedures, such as the covariance matching (CM) approach, SEM-based estimation provides parameter estimates of similar quality.
about the speaker
Fan Y. Wallentinis a Professor of Statistics at Uppsala University, Sweden. She received her Ph.D. in Statistics in 1997. She is a recipient of the Arnberg Prize from the Swedish Royal Academy of Sciences. Dr. Wallentin’s research program is on the theory and applications of latent variable modeling and other types of multivariate statistical analysis, particularly their applications in the social and behavioral sciences. She has published research articles in several leading statistics and psychometrics journals. She has taught courses related to latent variable modeling in Sweden, USA, China, and several European countries. She has broad experience in statistical consultation for researchers in social and behavioral sciences.