The field of Design of Experiments (DoE) has undergone a remarkable transformation over the past century, evolving from manually constructed designs to sophisticated computer-generated experimental plans.
Classical designs of this era were characterized by their elegant simplicity. Their construction relied on systematic rules: circular permutations and combinatorial principles that could be executed with just paper and pencil.
The 1970s marked a transformative period in experimental design, driven by the rapid growth of computing power and the emergence of algorithmic optimization methods.
The concept of optimality criteria became central to this revolution. Different criteria were developed to address specific experimental objectives.
New optimality criteria and algorithmic developments continue to expand the capabilities of computer-generated designs.
This convergence has created unprecedented flexibility in experimental design.
Modern innovations continue to emerge from both traditions. This suggests that the future of DoE will likely involve further synthesis of traditional principles with advanced computational methods.
Definitive Screening Designs (DSDs), Orthogonal Minimally Aliased Response Surface (OMARS) designs, and Orthogonal Main Effects Screening Designs for Mixed-Level Factors (OML designs) represent advances in classical design construction principles.
Current research explores the intersection of model-based and model-agnostic frameworks, promising even greater flexibility and efficiency in experimental design.
These developments enhance our ability to design efficient and effective experiments across diverse fields of application.