Covariate factors choice

[#1]

Hi,

Lets say I have a dataset of 100 measurements (pieces) where 4 material properties (parameters) were measured.

I want to include those 4 parameters as a covariate factors in custom DoE, together with 1 additional parameter, and I want to run 10 experiments. Thus, 10 pieces are chosen and additional parameter is varied between -1 and 1. 

Can somebody please explain to me how those 10 pieces/points are chosen? Are those the most optimal points in terms of covering the parameter space most efficiently, or is there any other metric? In terms of D-, I-, or A- optimality, how would those 10 points differ when choosing different optimality criterion?

Thanks!

[#2 · Accepted solution]

Everything is the same as when you use continuous factors, except that the coordinates are not randomly chosen within the bounds defined by the factor ranges. The coordinates are fixed. The coordinate exchange algorithm selects the optimal subset matched with your freely chosen continuous factor.

[#3]

Thanks.