Supermarket Application Analysis Based On The Association Rules Algorithm

With the rapid growth of economy,the operation scale of domestic brand chain supermarkets has been expanded currently and the management in the supermarket industry has been more scientific.A vast amount of data,commonly accumulated by modern chain supermarket in daily manage.However,has been poorly utilized by the operators,which results in the ignorance of the implicit value of business information.Therefore,the crucial significance lies in how to use the business data effectively,in how to collect knowledge or create value base on the large volumes of business data on the modern supermarket.A supermarket not only faces directly competition from other brands of supermarkets near the surrounding area,but also challenges from online retail,a new retail model.The core problems that the modern supermarkets how to adopt the appropriate data minning to analyze the business datas,discover the commercial law,create the economic value.Association rules serve as an important technology in the data mining,which has a wide range of applications.A certain amount of studies and applications have been carried out in the supermarket industry.Furthermore,hidden laws behind business datas found through the association rules,help operators deal with several problems in product display and promotion strategies.First of all,this thesis introduces relevant theories of data mining technologies,analyzes the characteristics of supermarket industry and lists the data mining techniques frequently used in the retail industry and the mining process.Secondly,the thesis elaborates data preprocessing technique in detail and indicates the necessity and popular methods of data preprocessing technique.Then deep analysis and research are launched in the association rules and its classic Apriori algorithm to present a comprehensive classification and summary of the association rules,to introduce the Apriori algorithm,to analyze the performance issues through examples of Apriori algorithm and to discuss a variety of improved methods of Apriori algorithm.Based on this,a better algorithm about Boolean matrix algorithm is proposed aiming at handling major bottlenecks of Apriori algorithm.In the aspect of compression matrix and the optimization of frequent item sets,improved algorithm mainly reduces the loop iteration and improves perfonnance.In addition,the thesis adds some new evaluation criteria to optimize the selection of mining results,focusing on the problems in the traditional evaluation model of"support-confidence" of associated rules.Finally,based on actual transaction data in a mid-size supermarket in the local area,a small practice system is developed,through data mining is carries out and association rules are discovered,for combining the relevant theories presented above,making Apriori algorithm and improved algorithm as technology carriers and"support-confidence-related interests" as an evaluation criterion.What's more,the system provides supports in business decision in terms of evaluation of mining results,identification of the valuable information among merchandise,shelf distribution,product display,cross-selling and promotion.