Studies On Mathematical Methods Of Transforming Multi-value Data Into Single Qualitative Symbols

Formal Concept Analysis was a theory developed by the German mathematicianProfessor Wille in 1982, which based on mathematical concepts and the concept lattice asthe core of the data structure. Formal Concept Analysis can be used as a data miningalgorithm for data analysis and knowledge processing. Its concept lattice structure modeldescribes essentially the link between objects and features, indicating that thegeneralization of cases of relations between the concept lattice and corresponding Hassediagram is data visualization.So it can be used for data analysis and knowledgeprocessing.The data can only be handled by this algorithm is discrete attribute data,but actauallymost of date is continuous attribute values, so it is necessary to discretize the continuousattribute data. Discretization of continuous attributes is to find discrete divised points, therange of the property is divided into discrete intervals, and Representing the attributevalue in each subinterval using these different symbols or integer values. Discretization ofcontinuous attributes belong to data preprocessing ,which is an important researchdirection in rough set theory.The main work of the paper is as follows：Firstly the paper introduces the history of the development on the discretization ofcontinuous attributes at home and abroad, and a brief description of the purpose andsignificance of the thesis topics. Then it summarizes the methods of using the ruler in theform of background substitution of the multi-valued attribute data into the qualitativeproperties of the data,which paves the way for the develop paper.Secondly, the paper classifies the discretization of continuous attributesSystematically.And on this basis,detailedly summarize of the advantages anddisadvantages on various types of discrete algorithms and the correspondingimprovements. It is proved that there is not a generic kind of discretization algorithms,usually we find the appropriate algorithm according to the different needs of differenttypes of data sets and data, as far as possible the results of the discretization optimal. Finally, it designs experiment using knowledge of Formal Concept Analysis , we getthe experimental results by data that treated by discretization and scale substitution.Thenwe can data analysis in order to achieve the purpose of the experiment.