A data warehouse for designed experiments (Q1584184)

Many organizations now record data from series of experiments. These data will often have been very costly to collect, but they can provide a valuable resource for future research. For example, in agriculture, they may be used to construct decision support systems or to study such issues as sustainability or climate change. In industry they may be used as part of the ongoing monitoring of a process, or to suggest aspects for future improvement. In this paper, methods are described for the efficient representation of designed experiments within information systems. Details to be stored include the physical properties of the units, the underlying structure of the design, the treatments studied in the experiment and their relation to the underlying structure. This relationship can succinctly be described by storing the generator used to construct the design. Possibilities include design keys, alpha arrays and initial blocks of cyclic designs. The same methods that are used to store the generators can be used to form databases containing repertoires of potential designs. Thus they also enable systems to be written to support the experimental process. A particularly convenient framework for such systems is provided by the visual statistical environment generated by the EU STABLE project.