Linear response surface analysis and some applications

Abstract

In general response surface analysis refers to the modeling of experimental or observational data. Computer technology and graphical techniques obviously play a significant role in the representations of these models. A well known traditional application is the use of hill climbing techniques and the fitting of models to experimental data by statisticians. In this paper we restrict attention to the analysis of linear response surfaces. In contrast to methods that explore the behaviour of nonlinear response functions in unconstrained situations we 1ook at the behaviour of linear response functions over the convex hull defined by observational data. The use of linear programming and related techniques play a major role in this type of analysis. A number of researchers have contributed towards these modeling techniques. These contributions are discussed and some advances that have been made with decision support systems based on this methodology are illustrated.