rsm package provides functions useful for designing and analyzing
experiments that are done sequentially in hopes of optimizing a response surface.
ccd can generate (and randomize) a central-composite
design; it allows the user to specify an aliasing or fractional blocking structure.
bbd generates and randomizes a Box-Behnken design.
ccd.pick is useful for identifying good parameter choices
in central-composite designs. Functions
djoin are also provided to build-up designs from individual blocks. The function
varfcn allows the experimenter to examine the predictive capabilities of a design before collecting data.
rsm is an enhancement of
lm that provides
for additional analyses peculiar to response surfaces. It requires a model formula
that contains a call to
SO to specify a first- or
second-order model. Once the model is fitted, the
function may be used to obtain the direction of steepest ascent (or descent).
canonical.path is an alternative to
steepest for second-order
In RSM methods, appropriate coding of data is
important not only for numerical stability, but for proper scaling
of results; the function
coded.data and its relatives facilitate
this coding requirement.
Finally, a few more functions are provided that may be useful beyond response-surface applications.
image.lm aids in visualizing a response surface,
or of any other
lm object where a surface is fitted.
recovers the data used in a
lm call, but unlike
polynomials, factors, etc. are expanded.
For more information and examples, use vignette("rsm") and vignette("rs-illus"). Additionally, vignette("rsm-plots") provides some illustrations of the graphics functions.
Box, GEP, Hunter, JS, and Hunter, WG (2005) Statistics for Experimenters (2nd ed.), Wiley-Interscience.
Lenth RV (2009) ``Response-Surface Methods in R, Using rsm'', Journal of Statistical Software, 32(7), 1--17. https://www.jstatsoft.org/v32/i07/.
Myers, RH, Montgomery, DC, and Anderson-Cook, CM (2009), Response Surface Methodology (3rd ed.), Wiley.