STATISTICS SEMINAR::Latent variable analysis for nonlinear structural modeling

GUEST LECTURER
Dr. Yasuo Amemiya

AFFILIATION
IBM T. J. Watson Research Center, NY

ABSTRACT
Latent variable analysis is a multivariate statistical method that has been
useful in various applied fields. The method is now utilized in problems
arising from business intelligence, system engineering, and public health,
in addition to those in behavioral/social sciences. In this talk, a
general class of latent variable statistical models is considered in the
framework of structural equation system. In such a system, observed
variables are assumed to be related to latent factors, and relationships
among the unobservable factors are to be studied. Recent results in
handling various kinds of nonlinearity are presented. Model fitting,
estimation, and inference procedures are discussed.

DATE & TIME
Thursday, September 09, 2004 -- 12:00 PM

DURATION
1 hour

LOCATION
228 ISyE Main Building

CONTACT PERSON
roshan@isye.gatech.edu