A comparison of bayes estimators and constrained least square estimators

Abstract

The mathematical properties of the Bayes estimator (BE) and the mixed estimator (constrained least square estimator (CLS)) for stochastic constraints, are compared. Conditions under which they have the same mathematical form are given. The mean square error (MSE) of the two kinds of estimators are compared. Ridge type estimators are shown to be special cases of both BE and CLS.

Keywords:

• Stochastic prior information
• biased prior information
• ridge and contraction type of estimator
• incorrect assumptions