Propuesta para aumentar los puntos experimentales en diseños D-óptimos bayesianos

One of the most frequent used criteria to obtain optimal designs is D-optimality designs, which provides experimental points where the volume of confidence ellipsoid associated to the vector of parameters in the proposed model is minimized. Unlike the classical D-optimal design, the Bayesian D-optimal design does not necessarily have as many support points as the model parameters. This article considers the case where D-optimal design averaged by a specific a priori has as many support points as the number of parameters of the model. This situation may not be as favorable when the model is not specified with complete certainty, since it would not be possible to conduct tests due to lack of fitness for the model. This article proposes a methodology that allows increasing the number of support points of the design in order that, with the resulting design, goodness of fitness test can be applied. Finally, the methodology is exemplified with an exponential model.

• Autores: Tellez, Cristian Fernando; López Ríos, Víctor Ignacio
• Fecha de publicación: 2013
• Institución: Universidad Santo Tomás
• Repositorio: Repositorio Institucional USTA
• Idioma: spa
• OAI Identifier: oai:repository.usta.edu.co:11634/39579
• Acceso en línea: https://revistas.usantotomas.edu.co/index.php/estadistica/article/view/1098; http://hdl.handle.net/11634/39579
• Palabra clave: D-optimalidad bayesiano; bondad de ajuste; D-eficiencia; incremento puntos experimentales; diseños óptimos
• License: http://purl.org/coar/access_right/c_abf2