Three Uncertainty Measures in Neighborhood Systems

Abstract

Uncertainty measures are important. In neighborhood systems, neighborhood accuracy, neighborhood roughness and neighborhood approximation accuracy can be used for evaluating uncertainty. However, they can not provide enough information in some situations. The paper depends on knowledge granulation to propose three uncertainty measures. Firstly, in neighborhood systems, neighborhood accuracy, neighborhood roughness and neighborhood approximation accuracy based on knowledge granulation are defined; Secondly, some important properties of the three new measures are studied, such as granulation monotonicity; Finally, two examples are designed to make illustration and comparison, and the three new measures achieve better performance. Neighborhood accuracy, neighborhood roughness and neighborhood approximation accuracy based on knowledge granulation are reasonable and effective.