Discretization method based on binary ant colony and variable precision rough set

Abstract

Discretization plays an important role in data pre-processing and has been used in fields such as artificial intelligence and data mining. A new discretization algorithm based on binary ant colony and variable precision rough set is proposed in this paper. Binary ant network is first built using candidate breakpoints, then global optimal breakpoints is searched in the network. The fitness function is established by the number of breakpoints and approximation classification accuracy of variable precision rough set. This method is compared with other algorithms using C4.5 classifier on the WEKA (Waikato Environment for Knowledge Analysis) platform and seven UCI data sets. The results indicate that the proposed method performs well.