Data Reduction Via Decision Trees

Abstract

Due to the wide availability of huge and enormous data that are stored in different repository systems (i.e. databases and data warehouses, etc…) have risen the needs to develop intelligent tools and techniques to infer or extract knowledge from such data. The answer to such need appeared in the field of Data Mining (DM), which is concerned with the discovery of knowledge from huge amounts of data. Eventually, these huge quantities of data slow the discovery process. Therefore, data preprocessing (i.e. data reduction) is an essential step to reduce the data size and in turn speeds up the discovery process.  The purpose of this paper is to employ decision trees in the context of data reduction and the ID3 (Iterative Dichotomiser3) method is the decision tree algorithm. The application of this idea was used in the test bed software system's implementation. The system is put to the test using well-known data sets to assess the viability of this method.