Information Technology Reference
In-Depth Information
Figure 7.2
Three-dimensional CA biplot for the 2007/08 crime contingency table.
Proportional to Pearson residuals in deviation form
R
−
1
/
2
C
−
1
/
2
; constructed from
(
X
−
E
)
1
/
2
1
/
2
.
U
and
V
keeping all other arguments at their defaults. In order to appreciate this three-dimensional
biplot we suggest the reader construct it using function
cabipl
andtheninteractively
rotate the biplot using the left mouse button and zooming in or out using the right
mouse button. Here we will discuss the two-dimensional biplot (given in Figure 7.3) in
more detail, referring to the one- and three-dimensional CA biplots by way of compar-
ison only.
The two-dimensional biplot constructed from the first two columns of
U
1
/
2
and
2
is given in Figure 7.3 with both rows and columns in the form of points. Figure 7.4
is the same biplot but now, according to our aim of constructing biplots to be used
analogously to ordinary scatterplots, with columns represented by calibrated axes. In
Figure 7.4 we have retained the points representing the columns to illustrate that these
points (in red) appear exactly on the corresponding axes. This biplot is the result of the
R call
1
/
V
cabipl(X = as.matrix(SACrime08.data), axis.col = "grey",
marker.col = "black", offset = c(2, 2, 0.5, 0.5),
offset.m = rep(-0.2,14))
