Construction D-optimal designs of experiments for linear multiple regression with heteroscedastic observations

Abstract

In article the problem of construction exact D-optimal designs of experiments for linear multiple regression in a case when variance of errors of observations depend on a point in which is made is investigated. Class of functions which describe change variance of heteroscedastic observations is defined for which it is possible construct D-optimal continues designs of experiments. For linear multiple regression with three factors it is constructed five different types of D-optimal continues designs of experiments with heteroscedastic observations. For each of these types the own class of functions describing change variance of observations is defined.

Keywords

• multiple regressions
• homoscedastic observations
• heteroscedastic observations
• saturated optimal designs