Research On Visual Mining Based On High Utility Pattern

To take the full advantage of the value in resources information, the technology of data mining has been emerged as the time require. However, it is private of users for mining process on traditional mining. Algorithm just deals with the data set that users given, and then output the result, which can be only acknowledged by the professionals studied on data mining, and users are also lack of trust on mining result. Given to the intuitive and figurative of graphical, image and colors, which can make up the defection of traditional mining by the combination of data mining and visualizing: on one hand, it enhances the participation of users, on the other hand, it emphasize the understand and usage of mining result. The technology makes the mining more intuitive, the mining process and result can effectively improve the understanding and credibility.High utility mining pattern is an important research direction of data mining. It is difficult to reflect the difference on items based on the support of frequent pattern mining. High utility pattern mining can make up this shortage and gradually become the hot spot in area. To meet the scenario of data stream, an algorithm based on transaction-sensitive sliding window, called MHUIDS (Mine High Utility Item sets over Data Stream), which for mining high utility item sets over data stream is proposed. The algorithm contains four parts:data set representation,window initialization, window sliding and high utility mining. Firstly, a tree structure, called the High Transaction-Weighted Utilization Itemset Tree (HTWUI-Tree), is introduced based on binary vector. Then, initialization and sliding algorithms of transaction-sensitive sliding window are described respectively. Finally, pruning strategies and mining algorithm are proposed. Experimental results show that MHUIDS algorithm is efficient and consumes low storage cost.This paper studies the possibility of high utility mining applied to the visual method. First, make the data set as an input, mining the high utility itemset by MHUIDS, then display the result by the nested rectangles. Last, drawing the statistic characteristics of the HTWUI or HUI in each sliding window. The results not only improve the accuracy of classification of product shelf display, but also provide the theoretical model and the scientific basis for the sales profit.