Practical Applications of Data Mining

82 Chapter 3 ClassifiCation learning

rules are the ultimate representation of the knowledge used in all algorithms

for classification learning, whereas decision trees are only used in the “divide

and conquer” and “partial decision tree” algorithms as an intermediate tool.

3.3 SEPARATE-AND-CONQUER APPROACH

The “separate and conquer” or “covering” technique takes each class and

creates rules that cover as many instances of this class as possible while

excluding as few instances of other classes as possible. This approach

examines only one class at a time. For each class, it builds a rule by selecting

Tree 1

Tree 2

24

Membranous Hemi-Sclerotized ...