Introduction
Design of experiments (DOE) is a part of statistics that provides tools for efficient experimentation. Although the subject started in an agricultural context, it is nowadays being applied in many areas, both in science and in industry. A model-oriented view, where some knowledge about the form of the data-generating process is assumed, naturally leads to the so-called optimum experimental design (OED).
Due to increasing competition, DOE has become crucial for modern industry, especially for product development. Since different industrial applications of DOE (especially those in pharmaceutical industries) may exhibit very special and varying characteristics, this model oriented approach with its tailor-made solutions is of advantage.
Research Group
Just few scientists in Spain are focusing on the optimum experimental design. It could be fairly stated that the research groups in the University of Almería , University of Salamanca and the recently created at the University of Castilla-La Mancha are almost the only ones doing research in optimum designs theory applied to regression models.
The conections between these groups have given way to different results as several PhD thesis and a number of publications. The association of these three groups has obtained some public funds from the Spanish Agency: Ministerio de Educación y Ciencia as well as from the corresponding Regional Administrations and Universities. The groups have also worked on several projects regarding different topics of Statistics, Probability and Operations Research.
The "Optimum Designs Model Oriented Design and Analysis Conference", mODa 8, took place in June 2007, 4th-8th, Almagro (Ciudad Real, Spain); the Local Organising Committee was composed by the members of the Optimum Experimental Design group.
Research Lines
Some of the current research lines of the group could be summarize on obtaining Optimal Designs for:
- Nonlinear models.
- Pharmacokinetic models.
- Multifactorial models.
- Compartmental models.
- Survival analysis and censored variables.
