Research Of Profit-Weighted Data Mining Algorithm Based On Association Rules And Its Application In Business

In recent years, with the huge amount of data accumulated and competition intensified, decision-makers need to search the information fast and timely from these massive data to make higher economy, which makes the application of data mining more and more widely. Business intelligence based on data mining is also increasingly important, due to their ability to help enterprise managers timely and correct to make decisions, or to resolve the problem.In data mining technology, association rule mining is widely used, of which Apriori algorithm is the classical algorithm. In the transaction database, Apriori algorithm is on the default premise that all the items have the same value in data mining process. But in reality, the different items which we call commodities that also have the different importance. In other words, different commodities bring the different profit to merchants. So this paper from a point of view on profit to help companies increase profit with data mining methods.This paper first introduces the concept of Business Intelligence, data mining and the choice of data mining tools. Then it describes the association rule mining techniques in detail. Because the shortcoming of traditional association rule mining algorithm, this paper presents a profit-weighted idea of different commodity items. Before data mining of association rules, the original transaction data firstly is conducted to data pre-processing through the application of this weight of commodity items. Thus this method reduces the scale of data mining and raises the efficiency of mining. For the support and confidence of weighted Apriori algorithm is no longer applicable, it gives the definition of support and confidence which base on weight. According to the weighted ideas, Boolean association rule mining algorithm based on profit-weighted is proposed, called LRJQ algorithm, which on the basis of improving on the original Apriori algorithm. This algorithm is compared with Apriori algorithm by experiments, and it has been proven to be effective. Finally using platform of SQL SERVER 2005 Business Intelligence, LRJQ algorithm is applied to cross-selling of goods system. In this step, first the dimension of this system is designed. And through extraction,transformation and loading of SSIS, transaction records which satisfy the requirements of associated mining are extract to the data warehouse. Then, the mining model is completed, after LRJQ algorithm plug-in is designed through SSAS. Finally, this article demonstrates the results of data mining. According to the results of mining can be more reasonable to conduct business supermarket layout, and to improve the sales effectiveness of businesses.