This report is aimed at readers with basic knowledge of logics and discrete mathematics.
Chapter 2 is about the concept of data mining with an emphasis on uncertainty handling. First we take a general look at data mining, and then a few ways of representing knowledge is given. Problems and challenges in data mining is discussed next, after which we look at how one can evaluate results given by a data mining system. Finally some methods for data mining is discussed.
Chapter 3 gives an introduction to the theory behind Rough Sets. Readers familiar with the theory might skip this chapter. Chapter 4 discusses an algorithm for creating so called default rules using Rough Sets theory. By generation of default rules, one get a much richer rule base than by just generating deterministic rules.
In Chapter 5 we look at classification based on rules. First we discuss different ways of using probabilistic rules from a theoretical viewpoint before we give details about the system implemented which use the default rules from the algorithm in Chapter 4.
A summary of the test results are presented in Chapter 6, before state of the art and related work are discussed in Chapter 7. Much used data mining systems and some particular systems using Rough Sets are discussed. Based on the performance of our system compared to others, conclusions are drawn in Chapter 8 and future work is outlined.
In the appendix there is a user's guide to the implemented system, and a description of the file formats used. In addition, the appendix contains test results, implementation documentation and our source code.