Page 147 -

134 4 Statistical Classification

dimensionality ratio at both levels, and therefore we obtain reliable estimates of the
errors with narrow 95% confidence intervals (less than 2% for the first level and
about +3% for the car vs. {fad, mas, gla) level).

Figure 4.41. Hierarchical tree classifier for the breast tissue data with percentages
of correct classifications and decision functions used at each node. Left branch =
"Yes"; right branch = "No".

DISCR. Rows: Observed classific.
ANAL. Columns: Predicted classific.
car con adi fad+
Group m l p = l981p=. I&=. 20&= 1621
52.4 11 0 0 10
con 0 9 2 3
0 1 2 1 0
1 0 0 4 8
10 2 3 6 1
m
Figure 4.42. Classification matrix of four classes of breast tissue using three
features and linear discriminants. Class fad+ is actually the class set uad, mas,
gla 1 -

For comparison purposes the same 4 classes discrimination was carried out with
only one linear classifier using the same three features 10, AREA-DA and IPMAX
as in the hierarchical approach. Figure 4.42 shows the classification matrix. Given