Data Mining: Practical Machine Learning Tools and Techniques, Second Edition

(Brent) #1

10.2 EXPLORING THE EXPLORER 385


Figure 10.11 shows the output. The pruned model tree is simply a decision
stump with a split on the MMAXattribute and two linear models, one for each
leaf. Both models involve a nominal attribute,vendor,as well as some numeric
ones. The expression vendor=adviser,sperry,amdahlis interpreted as follows: if
vendoris either adviser, sperry,or amdahl,then substitute 1; otherwise, substi-
tute 0. The description of the model tree is followed by several figures that
measure its performance. These are derived from the test option chosen in
Figure 10.10(b), 10-fold cross-validation (not stratified, because that doesn’t


(a)


(b)


Figure 10.10Processing the CPU performance data with M5¢.

Free download pdf