The
wafer
data frame has 18 rows and 13 columns, of which 8 contain
factors, 4 contain responses, and one is the auxiliary variable
N
.
It is a design object based on an orthogonal-array design for an
experiment in which
two integrated circuit wafers were made for each
combination of factors, and on each wafer the pre- and post-etch line
widths were measured five times. The response
variables are the mean and deviance of the measurements.
As 3 of the wafers were broken, the auxiliary variable
N
gives the
number of measurements actually made.
This data frame contains the following columns:
"category"
, levels are 2 < 2.5
204,90
,
206,90
, and
204,105
"category"
, levels are low < normal < high.
"category"
, levels are 20 < 30 < 40.
1
,
2
, and
3
.
"category"
,
levels are `-20% < normal < +20%'.
"category"
,
levels are ` 30 < 45 < 60'.
"category"
,
levels are ` 13.2 < 14.5 < 15.8'.
Phadke, M.S., Kackar, R.N., Speene, D.V., and Grieco, M.J. (1983)
Off-line Quality Control in Integrated Circuit Fabrication Using
Experimental Design. Bell System Technical Journal
Vol. 62, pp. 1273--309.
John M. Chambers and Trevor J. Hastie, (eds.) Statistical Models in S,
Wadsworth and Brooks, Pacific Grove, CA 1992, pg. 147.
wpm <- wafer[,c(1:9)] waov1 <- aov(pre.mean ~ . , wpm)