Bridging Theory to Clicks: Two-Level Designs with Treatment Cardinality Constraints in JMP (2026-EU-30MP-2792)

Modern experimental designs often face the so-called treatment cardinality constraint, which is the constraint on the number of included factors in each treatment. Experiments with such constraints are commonly encountered in engineering simulation, AI system tuning, and large-scale system verification.

It calls for the development of adequate designs to enable statistical efficiency for modeling and analysis within feasible constraints. We developed two-level designs tailored to cardinality constrained settings that balance factor activity, maintain good projection properties in low-dimensional subspaces, and reduce aliasing to support interpretable modeling. Our construction builds on balanced incomplete block design (BIBD) and uses an efficient genetic algorithm (GA) to explore the constrained design space.

To enable practical adoption, we used a JMP add-in that leverages existing JMP DOE and evaluation tools, while incorporating our construction algorithm for end-to-end design generation and assessment. The add-in automates checks of projection quality and alias structure. Numerical examples demonstrate improved estimation accuracy and clearer factor effects under treatment cardinality constraints.