Research On Multigranulation Rough Set Model And Its Application In Interval Value Ordered Information Systems

Big data plays an important role in every intelligent industry and field.How to obtain valuable information from the constantly updated massive data is an urgent problem that needs to be solved.As an important tool to deal with uncertain knowledge,rough set has been widely used in data mining,machine learning and other fields.However,the traditional rough set model requires the data to be accurate and complete,there are some limitations for the processing of interval value data.As an important extension type of single-valued information system,interval-valued ordered information system is an important model for processing uncertain data.In this thesis,the problem of multigranularity rough set modeling and object sequencing is studied based on interval-valued ordered information system.The main research contents are summarized as follows:(1)In interval value ordered information systems,Constructing a single granularity rough set model based on α advantageous relationships and studying its basic properties.When comparing the advantages and disadvantages of any two objects in an interval value ordered information system,not only the size relationship of the midpoint of the interval value is considered,but also the size relationship of the interval value radius is considered.It effectively solves the problem of two interval values that cannot be compared.Then,the single granularity rough set model is extended to a multi granularity perspective,the multi granularity rough set model is reconstructed,and decision theory is introduced into the multi granularity rough set model.Finally,the feasibility of the multi granularity rough set model is verified through comparison.(2)In the interval value ordering system,the traditional object sorting method is extended to a multi granularity perspective,and based on this object sorting method,combined with a multiple linear regression model,a predictive object sorting model is established.When a new object is added to the order information system,this model can be used to predict the sorting score of the new object.Based on the sorting score,the sorting of the new object in the order information can be obtained.And the effectiveness of this method has been demonstrated through experiments.(3)Empirical analysis of the smartphone market,using the theories and methods proposed in this article to analyze the smartphone market.Firstly,the best-selling attribute in the decision attribute is used as a concept set,and the upper and lower approximations of multi granularity rough sets are used to describe the concept set.The smartphone brands that are certain to be best-selling and not necessarily best-selling under different consumer preferences are obtained.Then,the best-selling level of smartphone brands was ranked under all preferences,and they were classified by whether they produced or sold.