The Application Of Improved Association Rule Algorithm In The Purchasing Data Mining

With the information society coming, the data storage capacity is rising dramatically. However, decision-makers distilled valuable information from large quantity of true and valuable data is quite difficult. Because the date is disorderly and one dimension. Faced with this challenge, data mining comes into being. With improving computer capability, degressing costs, and the succeed manage of data management technology, data mining is used in some decision-making system more and more.There are many data mining research directions, and association rule mining is the most active one in these directions, it reflects the significative association or correlative relation between projects in a lot of data. The most classical association rules mining algorithm is Apriori algorithm. Whereas Apriori algorithm in mining frequent itemsets need to produce a mass of candidate itemsets and scan database time after time, so time and space complexity is too high. In allusion to this limitation, how to enhance the efficiency of mining algorithm becomes the core issue of association rules mining research. This thesis researches the association rules mining algorithm deeply, proposes a sort of improved algorithm, then applies this algorithm to the data mining of Shanghai Volkswagen Company purchasing information, finally gains some association rules which can provide reference for decision-making.My research works mainly focus on three aspects as follow:1. Theory introduce. This thesis expatiates on the basic theory of data mining and association rules. Through theoretical introduction can establish a foundation for algorithm research and system applications.2. Algorithm introduce and improvement. This thesis deeply analyzes the classical algorithm of association rules Apriori on the basis of theoretical understanding. In allusion to the characteristics of association rules mining in relational database, this thesis proposes a sort of new algorithm called Coding-Apriori based on local coding, introduces its theory and process form various aspects and demonstrates its implementation steps. Lastly, comparing Coding-Apriori and Apriori by some experiments, validating the validity and advantages of the new algorithm.3. System application. This thesis designs and develops a simple data mining system for Shanghai Volkswagen Company purchasing information database. The system can selectively use the classical Apriori algorithm and the newly proposed Coding-Apriori to mining association rules. This part not only validates the research and improvement on algorithm practically, but also exploits the application fields of data mining, achieves the combination of theoretical research and practical application.