Adaptive DOE technique is able to distribute the design points equally throughout the design.
It samples the space with a set of design points generated in such a way so as to
maximize the minimum distance between neighboring points. The technique can be seeded with any
arbitrary set of previous design points. It also allows the user to select clusters of factors
that are modified in parallel. A user should put a set of coupled factors in a single cluster.
By default all factors are placed in a single cluster. Cluster definition can be defined based on
prior physical insight or based on the results of an orthogonal DOE study (e.g. Box Behnken) to
identify coupled or uncoupled terms. The main effects plot can be used to identify the contribution
of the Cluster variables. If the contribution is low, the Cluster can be dropped by setting the
Cluster to zero (0). At any time in future iterations, the Cluster can be activated again.
Following images shows the example where new points generated using arbitrary set of matrix
uniformly distributed into the design space with each iteration.
The following figure shows the sample locations for three factors.
