Portal de Periódicos da CAPES

Mixture designs for a model linear in the proportions

1970; Oxford University Press; Volume: 57; Issue: 2 Linguagem: Inglês

10.1093/biomet/57.2.329

1464-3510

Niels G. Becker,

Optimal Experimental Design Methods

When the predictor variables of a regression model are the n proportions of a mixture, the experimental region is contained in an (n-1)-dimensional regular simplex. Suppose a model linear in the proportions is to be fitted to observed responses. Designs are considered which minimize the mean squared diviation averaged over the simplex, when the true model is taken to be a quadratic in the proportions.