Choosing the Weighting Constant in Compound Optimal Designs: cD-optimal Designs

Two alternative methods for choosing the weighting constant in cD-optimal designs are proposed. The methods are based on power hypothesis testing associated with both the significance of model parameters under study as the significance of a nonlinear function of interest. The two methodologies and the existing methodology in the literature (efficiencies methodology) are described. With an example, the weighting constant is found with methodologies proposed and is compared with the methodology of efficiencies. cD-optimal designs with higher powers, small relative errors and even better than the optimal designs with the efficiencies methodology are obtained.

• Autores: Flórez-Restrepo, César; López-Ríos, Víctor Ignacio
• Fecha de publicación: 2013
• Institución: Universidad EAFIT
• Repositorio: Repositorio EAFIT
• Idioma: spa
• OAI Identifier: oai:repository.eafit.edu.co:10784/14426
• Acceso en línea: http://hdl.handle.net/10784/14426
• Palabra clave: Optimal Experimental Designs; Information Matrix; Efficiency; D-Optimality; C-Optimality; Compound Optimal Design; Power Test; Relative Error; Diseños Experimentales Óptimos; Matriz De Información; Eficiencia; D-Óptima; C-Óptima; Diseño Óptimo Compuesto; Prueba De Potencia; Error Relativo
• License: Copyright (c) 2013 César Flórez-Restrepo, Víctor Ignacio López-Ríos