Factorial Designs
• fullfact(n, [levels])—Returns an n factor full factorial design matrix with two levels per factor unless specified otherwise in the levels vector. If levels is entered as a scalar, the same number of levels is used for all factors.
• fractfact(n, r) or fractfact(“gen”)—Returns a 2–level fractional factorial design matrix generated from default generators for fractional factorials with n factors and 2n−r runs, or from user-defined generators specified in the string “gen”.
The algorithm used for generating the default generators is described in Understanding Industrial Designed Experiments, by Stephen R. Schmidt and Robert G. Launsby. The fractional factorial returned has the highest resolution at the given fraction r.
This function is limited to designs with n ≤ 26.
• plackettburman(nruns)—Returns a Plackett-Burman design matrix with nruns runs. The number of factors by definition is nruns − 1 and it cannot be set.
This function is limited to designs with 8 ≤ nruns ≤ 48, and nruns must be a multiple of 4.
The fullfact and fractfact functions use the ordering specified by NIST — the National Institute of Standards and Technology.
Arguments
• n is an integer specifying the number of factors.
• levels (optional) is an n element vector, or an integer ≥ 2, specifying the number of levels for each factor. By default, there are two levels per factor.
• r is an integer specifying the fraction power.
• "gen" is a string specifying the design generators in a fractional factorial design, such as "A, B, C=AB, D=AC". The factor labels must be sorted in alphabetical order. For the separators, you can use a space, a comma, a colon, or a semicolon.
• nruns is an integer specifying the number of runs.