A Method of Discretization of Continuous Attributes in Knowledge Discovery

Discretization of continuous attributes is an important preprocessing step in knowledge discovery. Most consecutive property boundaries are blurred. It is suitable for soft classification. This paper introduced the fuzzy set of continuous attribute values are transformed into fuzzy attribute values. We presents the discretization method based on fuzzy similarity clustering. The experiment shows the method of discretization is more effective and objective than others. It is a effective method in discretization of the continuous attribute.