Some applications of canonical moments in Fourier regression models
Holger Dette · Lecture notes-monograph series · 1998
This paper applies recent results on canonical moments for the determination of optimal designs for multivariate Fourier regression models.Optimal designs for discriminating between different Fourier regression models can be found explicitly.It is also demonstrated that these designs may be useful in orthogonal series estimation and for testing additivity in nonparametric regression.In contrast to many other optimality criteria for the trigonometric regression model, the discrimination designs are not necessarily uniformly distributed on equidistant points.