Information Technology Reference
In-Depth Information
CrJk
0
0.8
0.95
2
0.8
0.5
RAC
AtMr
0.5
0.5
2
1.4
CmRb
1.5
0.5
Gaut
1.2
1.5
AtMr
KZN
InAs
DrgR
2
CmRb
3
DrgR
1.5
WCpe
1
1
InAs
2
0.9
CmAs
Mrd
CmAs
BRs
0
1.1
0.5
0.8
0.5
Arsn
BNRs
1.5
Mpml
PubV
NWst
Rape
0.6
0.5
FrSt
AGBH
0
ECpe
Limp
1.5
1.5
2
0.5
Arsn
NCpe
1.2
0
1.5
Figure 7.10
Two-dimensional CA biplot of the 2007/08 crime data set. The contingency
ratio is approximated by plotting
R
−
1
/
2
U
1
/
2
1
/
2
and
C
−
1
/
2
V
with correctly adjusted
scales.
ca.variant = Chisq2Rows
and
ca.variant = Chisq2Cols
, respectively) are
given in Figures 7.14 and 7.15. These biplots were used to obtain predictions for
the chi-squared distance matrix. The two-dimensional approximation of
√
n
times the
chi-squared distance matrix is given in Table 7.17.
The distances in Table 7.17 were calculated as follows. The biplot in Figure 7.14
provides the approximation to the matrix
R
−
1
(
X
-
E
)
C
−
1
/
2
. This approximation was used
for calculating the between-row chi-squared distances precisely in the same way as the
actual distances were calculated from
R
−
1
(
X
-
E
)
C
−
1
/
2
using the function
Chisq.dist
.
Likewise, we calculated
√
n
times the chi-squared distance matrix between the
columns of our 2007/08 crime data set. This matrix is shown in Table 7.18. The
approximation
based
upon
the
two-dimensional
biplot
in
Figure
7.15
is
given
in
Table 7.19.