Designed to make the trade-offs visible
Full factorial designs
Explore combinations of factor levels with seeded randomized run order. These designs support estimating effects under suitable replication and independent measurements; the number of runs grows quickly with each factor.
Latin hypercube sampling
Distribute a fixed number of runs across continuous factor ranges. This space-filling design is exploratory; it does not by itself provide balanced replication or establish causation.
Factors and measurement context
Record numerical or categorical factors, units, intended bounds, a response, and an objective. Add controlled conditions and measurement procedures in the experiment notes.
Versioned design decisions
Preserve design configurations and factor definitions with an author and time. New versions keep earlier decisions available for comparison.
Randomization, independent replication, measurement validity, confounding, and practical feasibility still require scientific review. A generated matrix is a plan, not a validated experiment.