A pessimistic multi-granularity ball rough set model is developed as the counterpart to the optimistic model,and a variable multi-granularity ball rough set model is designed to address differ-ent decision-making problems.The relevant properties of this model are explored,and uncertainty measures associated with it are introduced.Three distinct positive region generation algorithms for multi-granular ball rough set models are proposed.These algorithms partition the data into granular balls by adjusting the purity parameter,effectively capturing the inherent relationships within the da-ta.Finally,experimental evaluations on eight UCI datasets confirm the feasibility and effectiveness of the proposed model.
维普
万方数据
