Full Factorial

In full factorial screens, all elements of the matrix of parameters are sampled

Incomplete factorial screening is a method of sampling parameter space evenly and efficiently.
Factor levels are chosen randomly and  then balanced to acheive uniform sampling.
All two-factor interactions are sampled as uniformly as possible.

Purely random sampling of all parameters, but it approximates incomplete factorial designs.
 

Sparse Matrix screens involve an intentional bias towards combinations of conditions that have worked previously.

© 1999-2005 Airlie J McCoy, University of Cambridge. All rights reserved.

Last updated: 7 June, 2005