The Exposition of Fuzzy Decision Trees and Their Application in Biology - Advancing Artificial Intelligence Through Biological Process Applications - page 387

Figure 6. Fuzzy decision tree for abalone age problem

L with truth level 0.730. For the other two paths
from the root node, SLW = M and SLW = H, the
largest of the values in the sets of subsethood
values are each below the minimum truth level
of 0.72, previously defined, which means further
checks for the lessening of ambiguity are made
for these two paths, see Table 6.
In Table 6, the G (SLW = ·) column of values
identifies the level of classification ambiguity with
each subset partition from the SLW root node (only
M and H considered). These values are important
since they benchmark the required value below
which a weighted average of classification ambi-
guity with each subset of partition G ( P | F ) must
be to augment a node to a particular path.

For the path defined by SLW = M, with G (SLW
= M) = 0.659, the subsequently evaluated lowest
G (· | SLW = M) value was found to be associated
with the condition attribute SKW (value in bold),
with G (SKW | SLW = M) = 0.658. It follows, the
path SLW = M is augmented with the attribute
SKW and the linguistic terms it is described by
(L, M and H). In the case of the path defined by
SLW = H, with G (SLW = H) = 0.624, the lowest
G (· | SLW = ·) value is found to be, G (SEX | SLW
= H) = 0.625. Hence, with G (SEX | SLW = H) =
0.625 > 0.624 G (SLW = H), there is no further
possible augmentation possible that would reduce
the considered ambiguity for that path, so it is a
leaf node with less than the desired truth level.