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

Most of the techniques in this book produce easily comprehensible descriptions

of the structural patterns in the data. Before looking at how these techniques

work, we have to see how structural patterns can be expressed. There are many

different ways for representing the patterns that can be discovered by machine

learning, and each one dictates the kind of technique that can be used to infer

that output structure from data. Once you understand how the output is

represented, you have come a long way toward understanding how it can be

generated.

We saw many examples of data mining in Chapter 1. In these cases the output

took the form of decision trees and classification rules, which are basic knowl-

edge representation styles that many machine learning methods use.Knowledge

is really too imposing a word for a decision tree or a collection of rules, and by

using it we don’t really mean to imply that these structures vie with the realkind

of knowledge that we carry in our heads: it’s just that we need some word to

refer to the structures that learning methods produce. There are more complex

varieties of rules that allow exceptions to be specified, and ones that can express

chapter 3

Output:

Knowledge Representation

61