Research Of Mining Frequent Patterns And Classification On Data Straems

In recent years, data streams were generated in a variety of applications, such assensor networks, stock analysis, network monitoring. Different from traditional data,the data stream is abounded, rapid, and continuous. Traditional data miningtechniques can not be applied directly to the data stream. Using the limited storagespace to handle data stream for rapid access to useful information bring out newopportunities and challenges to researches on data mining and its applications.Thispaper aimed at frequent pattern mining and classification algorithms for thehigh-speed data stream.First, we studied the traditional theory and the classical data mining algorithms,including frequent patterns mining algorithm Apriori, FP-Growth and classificationalgorithm ID3.Second, we learned the characteristics of the data stream and its models in whichthe sliding window model is the best to the real applications. Therefore, based on thetraditional static algorithms, we proposed and implemented some single-passalgorithms including frequent pattern mining algorithms SOA, SFP and classificationalgorithms SDT,SFPC under the sliding window model.Finally, we designed and implemented the mining platform Sminer with B/Sstructure where the algorithms proposed above were tested, and the experiments showthat they have a high level of accuracy and time efficiency. In addition, this paper also analyzed the applications of frequent pattern mining and classification in networkmonitoring.