robust.sn(robust.df, snratio="target", response)
"robust.design".
snratio can take the values:
"small" denotes the smaller the
better,
"large" denotes the larger the better, and
"target", the
default, denotes the closer to target the better (nominal-the-best).
robust.df for
computing the signal to noise ratios.
The default is all columns not "noise" or "control" factors in
robust.df.
robust.sn which
inherits from
data.frame. It contains the design points of the
control array used to generate the
robust.df, and the following quantities
for each response in the data frame.
.
(dot).
The summary statistics computed by default are
mean (
mean), standard deviation (
sd), mean(log(y)) (
meanl),
log(standard deviation(log(y))) (
lsdl).
These can then be used as responses to study behavior of the mean and
standard deviation, via
fac.aov. In addition, one of Taguchis
signal-to-noise ratio statistics is computed.
If
"target" is specified, then the nominal-is-best ratio is computed:
If
"smallest", the smaller-is-better ratio is computed:
If
"largest" is specified, the larger-is-better ratio is computed:
Taguchi, G. (1986).
Introduction to Quality Engineering: Designing Quality Into Products and Processes,
Tokyo: Asian Productivity Organization.
Phadke, M. (1989).
Quality Engineering using Robust Design,
AT&T Bell Laboratories, Prentice Hall, New Jersey.
mold.sn <- robust.sn(mold.df)