Parameter estimation from multiresponse data

Abstract

Abstract Multiresponse data are familiar outputs of experiments and processes involving multicomponent mixtures, multiple streams or multiple methods of observation. The resulting arrays of data exhibit various structures, including rectangular (no missing values), block‐rectangular, and irregular (missing values with no simple pattern). Methods for modeling such data are described, emphasizing parameter estimation strategy and available software. Experiences in multiresponse modeling are then reviewed for several chemical engineering problems.