Optimal Granularity Selection in Multi-label Decision Table

To solve the problem of granularity selection in multi-label decision classification, a global and object-based optimal granularity selection method is proposed. Firstly, the granularity partitioning of multi-label decision tables is analyzed based on multiple granularity levels and the granulation roughness measurement method of multi-label decision tables is introduced. Then the general optimal granularity selection method is discussed for coordinated decision tables and uncoordinated decision tables. Finally, in view of the limitation that the global optimal granularity selection cannot make every object reach the optimal granularity, and the uncertainty of some objects in the uncoordinated decision table regarding the classification of decision markers, the local optimal granularity selection method of objects is discussed, and the effectiveness of the method is verified by an example.