Decision Table Decomposition for Further Rule Induction

Abstract

Classification is one of the main issues of data mining. Knowledge hidden in the data can be discovered by induction of decision rules. However, with the increase in the size of the decision tables there is a need to decompose the problem. An appropriate solution to this problem may be hierarchical induction of decision rules. In this article the decomposition algorithm of decision tables containing multi-valued attributes has been presented. It has also been shown that efficient algorithms derived from logic synthesis may be applied to the hierarchical induction of decision tables. Experimental results have proven that by using presented methods one can achieve a considerable data compression and acceleration of calculations.