Menu.ccdTab1 {RcmdrPlugin.DoE}

Basic information for Central Composite Designs

Description

Basic information for Central Composite designs

Brief statistical background

Central composite designs can have three (alpha=1) or five levels in all factors and are useful for fitting second order polynomials to experimental data based on several quantitative response variables. They can be used in a process of sequential experimentation, experimenting first with the cube portion and later adding (a) star block(s).

For more information, look at the help file for function ccd.design from package DoE.wrapper or at ccd from package rsm.

Inputs on Tab Base Settings

name of design

must be a valid name. The design itself is created under this name in the R workspace.

Determine the cube portion

An existing 2-level fractional factorial, preferrably of at least resolution V, can be selected, or a new design for the cube portion can be generated;
if the number of center points is to be different between cube and star portion of the design, the cube portion must be generated with center points.

number of center points

The number of center points in the cube portion is ideally defined within the cube itself. In that case, the number given refers to the center points in the star portion. If the cube portion has no center points, a single number of center points is applied to both the cube and the star portion.
Alternatively, a separate number can be given to specify different numbers of center points for cube and star portion, e.g. 0, 5 for the unfortunate situation where the cube portion has already been run without any center points (not recommended!).)

alpha

determines how far inside (alpha<1) or outside (alpha>1) of the cube the star points are positioned. The special values orthogonal (default) or rotatable define calculated alpha values with interesting statistical properties. However, it must of course be checked whether these make practical sense. For example, if the orthogonal alpha turns out to be 2 but experimentation beyond 1.5 does not make sense, the number should be defined manually.

seed

can be provided for reproducing an existing design.

randomization

can be switched off in exceptional situations. Usually, one would not want to do that. Note that the checkbox here refers to the star portion only. For the cube portion, there is also such a checkbox.

Author(s)

Ulrike Groemping

References

~put references to the literature/web site here ~

See Also

See also FrF2 for the function defining the cube portion of the design ccd.augment for the function that augments the cube portion with the star points, also function ccd from package rsm for the function that underlies function ccd.augment.