Algorithms Of Frequent Pattern From Uncertain Data

The mining of frequent pattern is one of the most popular research areas in data mining. However, with the improvement of the software and hardware technology and the expansion of applicable range, various institutions and organizations have accumulated huge uncertain data. For instance, data from patients’diagnostic, data from sensor networks, information of satellite image and so on. Because of the uncertainty of data, the traditional mining algorithm for frequent pattern is not applicable to uncertain data.In this paper, I take the frequent pattern mining algorithm of uncertain data as our research object. I make use of possible world as data modal, analyze and summarize typical frequent pattern mining algorithms of uncertain data. Then I conclude the general approaches of the frequent pattern mining algorithm of uncertain data in the present research.The contributions of this paper include the following:●I research and analyze the typical frequent pattern mining algorithm of uncertain data deeply. And I conclude the general approach of it, that is, sampling and extension.●I put forward a kind of frequent pattern mining algorithm of uncertain data based on sampling. In this algorithm, we should do some trimming and sampling before mining the frequent patterns on the uncertain data set. According to the analysis, present merits and defects of the algorithm are presented by its detailed steps.●I put forward another kind of frequent pattern mining algorithm of uncertain data based on extension. This algorithm extends the tidset as well as the itemset search tree. The Tid that contains only one id field is extended to a new Tid that contains both id field and probability field. Then the extended itemset search tree is consisted of the new tidset. The extended tidset can describe uncertain data, and the extended itemset search tree is built to mine the frequent patterns. The UP-Eclat algorithm is proved to be efficient according to the experimentation.