The k-means method of discretization for continuous attribute in rough set theory
Zhe Chen · Journal of Liaoning Technical University · 2015
In order to discrete the continuous attribute before application of rough set theory, this paper utilized a discretization method based on the k-means algorithm, which discreted attributes of unsupervised clustering method into two categories. Four sets of data on UCI database were chosen for experiment to verify the performance of the proposed method. First step was to discrete the data, and then they were used to do attributes reduction through rough set and recognition by k NN(k=10) classifier classification algorithm. The results then was compared with other two discretization methods. The experimental results show that this method can improve the efficiency of discretization, reduce the complexity of the experiment, and effectively reduce the break points.