A New Method for Discretization of Continuous Attributes

Because traditional Rough Sets theory can only deal with discrete attributes, continuous attributes must be converted into discrete attributes before we cope with decision table. In this paper, a new algorithm of consistent discretization in decision table based on rough sets theory is presented for the continuous attributes. The discretization algorithm is performed by selecting the same decision attribute value from the discretized attributes,and classfy the unions as a class. This algorithm is effectively tested by an example. The method’s complexity is analysed and compared with some other discretization algorithms. Results show that using this approach can get reasonable intervals.