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)