Research on attributes discretization in target fusion syetem

A new method for discretization of continuous attributes is trying to be discussed in this paper based on the importance of attributes that finds solutions to overcome the limitation of the traditional rough sets. According to consistency degrees, grouping is an effective way to select candidate cut points, it also helps reducing the numbers of cut points. So the consistency of the decision-making system is maintained in the form of attribute discretization which permits the reduction of cut point numbers and the improvement of efficiency. Adopting variable precision rough information entropy as measuring criterion, it has a good tolerance to noise. Experiments show that the algorithm yields satisfy this reduction results.