The general Nature of Saturated Designs

08/09/2019

∙

In a full two-level factorial experiment the design matrix is a Hadamard matrix H. The OLS estimator of the full set of parameters β ( the mean, the main effects and interactions) is given by β̂ = 1/NH^TY.Thus the estimator of each parameter except the mean is a contrast. That is H^T1_N = < b m a t r i x >. In this paper we show this result not only holds for Hadamard matrices but also holds for any saturated design matrix D in the two-level factorial experiment set-up.

READ FULL TEXT

The general Nature of Saturated Designs

Matrix quadratic risk of orthogonally invariant estimators for a normal mean matrix

On The Study Of D-Optimal Saturated Designs For Mean, Main Effects and F_1-Two-Factor Interactions For 2^k-Factorial Experiments

The cost-free nature of optimally tuning Tikhonov regularizers and other ordered smoothers

Joint Estimation of Location and Scatter in Complex Elliptical Distributions: A robust semiparametric and computationally efficient R-estimator of the shape matrix

Efficiency Requires Adaptation

The finiteness conjecture holds in SL(2,Z>=0)^2

Estimating the number and effect sizes of non-null hypotheses

The general Nature of Saturated Designs

Related Research

Matrix quadratic risk of orthogonally invariant estimators for a normal mean matrix

On The Study Of D-Optimal Saturated Designs For Mean, Main Effects and F_1-Two-Factor Interactions For 2^k-Factorial Experiments

The cost-free nature of optimally tuning Tikhonov regularizers and other ordered smoothers

Joint Estimation of Location and Scatter in Complex Elliptical Distributions: A robust semiparametric and computationally efficient R-estimator of the shape matrix

Efficiency Requires Adaptation

The finiteness conjecture holds in SL(2,Z>=0)^2

Estimating the number and effect sizes of non-null hypotheses