Application of Data Decomposition to Incomplete Information Systems

Rafal Latkowski

Institute of Computer Science, Warsaw University
ul. Banacha 2, 02--097 Warsaw, Poland
rlatkows@mimuw.edu.pl

Abstract

Many developed classification methods and knowledge discovery software,
that were research subjects for years, suffer from the lack of
possibility to handle data with missing attribute values. To adapt
existing classification methods to incomplete information systems, we
propose a decomposition method that allows more appropriate
missing value attributes handling. The decomposition method
consists of two phases. In the first step data from original decision
table are partitioned into subsets. In the second step, knowledge
from those subsets, that in our case is classification hypothesis, is
combined to achieve a final classification based on a whole original
decision table. There were carried out some experiments in order to evaluate
the decomposition method.