Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer software > Program design,software engineering > Programming > Database theory and systems

Research on Quantitative Association Rules Based on Data Field

Author LiDanDan
Tutor MengHaiDong
School Inner Mongolia University of Science and Technology
Course Applied Computer Technology
Keywords Data mining Quantitative association rules Cluster analysis DataField
CLC TP311.13
Type Master's thesis
Year 2013
Downloads 10
Quotes 0
Download Dissertation

With the wide application of data collection tools, collection and accumulation ofmassive data made it explodes, and it is beyond the ability of people to understand andmaster. The traditional data analysis methods can’t meet the needs, so the data miningtechnology came into being, and it is widely used. Association rule mining is an importantbranch of data mining research for discovering interesting relations between the set ofproperties that exist in the database. Mining quantitative association rules is an importantresearch topic in association rules mining. It is more and more attended and concerned bythe data mining community, because of its widely application in many fields such asbusiness, production.The article analyses the current situation and methods of data mining research indomestic and foreign, and studies the mining of quantitative association rule under thecontext of rule mining research. The article Firstly introduces the basic concepts of datamining technology, mining object, mining tasks, the basic process and classification, andthen describes some of the basic theory, concepts of association rules, and the steps ofmining association rules, focuses on researching the classic association rule miningalgorithm Apriori algorithm and the problems of it using in mining quantitative associationrules. On this basis, the article uses the idea of data field, and combined with thecharacteristics among the data presented in the data field, proposes quantitative associationrules based on data field. This method compared with the previous quantitative associationrule mining method has the following improvements. Firstly, when mining of quantitativeassociation rules, each data in data set is mapped into data point having certain data energyin the data field. Each data point radio the surrounding and receives radiation energy ofother data in independently to reflect the different roles for data mining tasks; Secondly,the clustering part, on quantitative attributes overall clustering makes clustering attributerange more reasonable; Thirdly, combined with the characteristics of the data field andanalyze the advantages and disadvantages of the K-means clustering algorithm, using good and abandoning bad, while choice of clustering algorithm, thereby enhancing the effect ofclustering; Fourthly, in the calculation of the support and confidence in consideration ofeach data different reflects in data mining tasks, so gets a more reasonable description ofthe support and confidence of the rules.In order to verify the effectiveness of the algorithm programming algorithm in theMicrosoft Visual Studio2008development tools, and the association rules miningIris dataset, in bodyfat data set, the clinical hospital data set through the programming accord withactual knowledge of the related fields.

Related Dissertations
More Dissertations