Latent Regression with Constrained Parameters to Determine the Weight Coefficients in Summary Index Model.

The constructions and algorithms of summary index models to determine unknown weight coefficients using latent variable regression with constrained parameters are presented. The summary index model with a single category can be seen as a structural equations model with a latent variable without latent equations. This article gives a new suitable algorithm for it based on factor analysis, path analysis, and prescription constraint. Another summary index model is one in which all samples have been divided into multiple categories. This article also gives a suitable algorithm by alternating projection between two convex sets and prescription constraint. [ABSTRACT FROM AUTHOR]