Font Size: a A A

The Research And Application Of Decision Tree Based On Intuitionistic Fuzzy

Posted on:2020-03-05Degree:MasterType:Thesis
Country:ChinaCandidate:Y F LiFull Text:PDF
GTID:2370330575498489Subject:Information management
Abstract/Summary:PDF Full Text Request
Decision tree algorithm has been widely used due to its straightforward and reusable characteristic as one of the classic classification models of machine learning and data mining.However,the classic Decision Tree algorithm can't deal with classification problems in fuzzy environments.So based on fuzzy set theory,a Fuzzy Decision Tree(FDT)algorithm is proposed,which is an extension of classic Decision Tree algorithm under fuzzy sets.This algorithm is to process an originally disordered sample data in a fuzzy way and then generate a model that can be used for classification prediction.The Fuzzy Decision Tree algorithm takes advantage of the extensive application scenarios of decision trees and the high comprehensibility of knowledge representation itself,as well as its ability to deal with fuzzy information by fuzzy theory.FDT has been widely used in medical,transportation,finance and other fields.FDT is essentially a supervised data classification model.Typically,it uses membership to indicate the extent to which a sample belongs to a node.However,in the actual data classification process,not only this indicator should be taken into consideration,but also the extent of not belonging to a node.Obviously,FDT cannot fully describe such classification information.The proposed Intuitionistic Fuzzy Decision Trees(IFDT)solves this problem,but the existing IFDT algorithm only gives the idea of construction and lacks the complete construction process.Moreover,the method of calculating the intuitionistic fuzzy entropy does not satisfy the constraint in some cases.Therefore,this article proposes an Improved Intuitionistic Fuzzy Decision Trees(IIFDT),which modifies the method of calculating the intuitionistic fuzzy entropy of IFDT,and proposes the pruning method,rule extraction method and classification prediction method of IIFDT.IIFDT includes a complete construction process,which improves the calculation where the intuitionistic fuzzy entropy does not satisfy the constraint in IFDT.In addition,IIFDT was applied to the diagnosis of diabetes,and an IIFDT-based diabetes-assisted diagnosis system was developed based on this algorithm.The contents of research in this article mainly include the following three aspects:(1)Discuss the construction idea of IFDT,sorts out the related theories of FDT,summarize the extended attributes selection methods and pruning methods of different FDT,and combs the basic concepts related to intuitionistic fuzzy sets,thus solidly constructs the theoretical basis of IIFDT.(2)Build an IIFDT model.The process of IIFDT construction is deduced,including the fuzzy processing of attributes,the selection criteria of extended attributes,pruning processing of IIFDT,the extraction of IIFDT rules and the classification prediction of IIFDT.Then the IIFDT model is constructed by using a case study.Finally,the IIFDT and several other algorithms are compared and analyzed using three sets of public data sets,which proves the advantages of the algorithm.(3)Develop an IIFDT-based diabetes-assisted diagnostic system.This article designs and implements an IIFDT algorithm based diabetes-assisted diagnostic system through three steps including system requirements analysis,framework design and detailed design.This diagnostic system contains four modules,which are personnel management module,data processing module,diabetes model building module and diabetes assisted diagnosis module.Finally,through the comparative analysis of the predictive results,the feasibility of the application of IIFDT model in assistant diagnosis of diabetes is illustrated,and the system has certain application value.
Keywords/Search Tags:Fuzzy decision tree, intuitionistic fuzzy decision tree, improved intuitionistic fuzzy decision tree, medical diagnosis system
PDF Full Text Request
Related items