Research of Attributes Discretization Based on Granular Entropy Considering the Discrimination of Rough Set

Attributes discretization is an important and difficult step in the data pretreatment for data mining. In order to overcome the defect that row-column measure is affected by the position of attributes and breakpoints, an approach for attributes discretization based on granular entropy considering discrimination was proposed. The approach searches the optimal discretization breakpoints sets in information table according to the characters of granular entropy and equivalence relation of rough set. And moreover, the feasibility and validity of algorithm was showed by the testing data.