Research On Extraction Method Of Decision Rules Based On Incomplete Multi-scale Information

With the advent of the big data era,data with different types often have a multi-scale structure.The analysis and processing of data under different scales of attribute information has become an important method for processing uncertain information.However,the existing multiscale data analysis models are mostly established for complete information systems.The optimal scale selection and rule extraction methods carried out on incomplete information systems are only limited to a fixed scale and cannot reflect the ideological characteristics of multi-scale information fusion.Based on the above considerations,this paper studies the decision rule extraction method based on incomplete multi-scale information system.First,extend the complete information system to the incomplete information system,consider the overall generalization reduction of the incomplete multi-scale decision information system.By introducing the concepts of granularity tree and pruning,the overall generalization reduction,optimal scale and the k-scale attribute reduction has been studied in detail.Then,the generalization reduction of objects are studied from a local perspective,and more general decision rules can be obtained by this method.Secondly,the incomplete generalized multi-scale decision information system and scale combination are studied,and the definition and properties of rough approximation under different scale combinations are discussed.Then,the optimal scale combination selection method of the coordinated incomplete generalized multi-scale decision information system is proposed,including the selection method of the global optimal scale combination and the local optimal scale combination.Finally,use an example to illustrate,design related algorithms,and further give the realization methods and specific steps of attribute reduction and rule extraction.