What inspired this wish list request? Currently, it's possible to create a Space-Filling design with categorical factors. However, it's not possible to create a Space-Filling design with categorical factors when using a candidate set approach.
Creation of Space-Filling design from scratch (using "Algorithm Data" dataset) :
Creation of a Space-Filling design from the dataset using a Candidate Set approach, using "Load Design" function (no categorical factor allowed) :
What is the improvement you would like to see? I would like categorical factors to be handled by the platform consistently in both cases : being able to use them when creating a design from scratch and when creating a design from a candidate set approach.
Why is this idea important? The Fast Flexible Filling algorithm is excellent at providing interesting Space-Filling designs in presence of categorical factors, and may be one of the very few Space-Filling algorithms able to handle such factor type. Being able to use historical data or covariate data that imply categorical factors would be a key addition for using the candidate set approach more smoothly. I know the Space-Filling selection of points could be done separately for each level of categorical factors, but it is not very practical and may not be as efficient as if the whole dataset can be used.