next up previous contents
Next: Accuracy of the Rules: Up: Classification Previous: Classification

Methods for Using Existing Rules

To classify objects which has never been seen before, rules generated from a training set will be used. We will assume these rules are of the form

A=a D=d with probability p.
where A is a set of condition attributes and D is the decision attribute.

First, we will discuss how to use the probability or accuracy of the rules in the classification process, and then we will look at a more advanced method which uses a principle called weight of measure.




Helge Grenager Solheim
Sat May 4 03:30:02 MET DST 1996