A computer tool for MV-optimal designs in binary response models

V. M. Casero Alonso, J. López Fidalgo, B. Torsney

Among the optimality criteria the MV-optimality is a potentially difficult criterion because of its nondifferentiability at equal variance designs.
On the other hand, binary response models are used in many real applications.
These models can be seen as weighted linear regression models when the weight function has a finite integral.
A general review of the literature of optimal designs for these models is made in this contribution.
The methodology for obtaining MV-optimal designs in a compact interval design space [a,b] is given for several standard weight functions.
An important contribution of this work is a user-friendly computer tool based on Mathematica for computing this designs in a number of situations.
Some illustrative examples are provided showing a representation of MV-optimal designs in the Euclidean plane taking a and b as the axes.

Palabras clave: c-optimality, Equivalence theorem, Minimax criteria, EV-optimality, Optimal design, Software.

M08.1 Grupo de Diseño Óptimo de Experimentos
6 de septiembre de 2016  15:20
0.02 - Aula de proyectos 1

E. Delgado Márquez, M. Amo Salas, J. López Fidalgo, L. Filová

J. López Fidalgo, R. Martín Martín, M. L. M. Rodríguez Hernández

M. Amo Salas, J. López Fidalgo