Research On The Algorithm Of Fractal Clustering Data Stream

Data-stream consists of a series of ordinal coming, boundless, dynamic data.Real-world applications often generate huge amount of data streamside, which challenges efficient processing and mining due to its special characteristics. The disadvantages of traditional data mining algorithms in mining data-stream is indicated by many researchers, meanwhile, these disadvantages also promote the researches of improving of existing data mining algorithms and creating new data-stream mining algorithms.As a important problem in data stream mining,clustering technique employed in these applications should be effective in terms of space usage.This has received considerable attention in the past few years due to its research value and increasing amount of importantance in numerous applications.In the thesis, fractal and some classical algorithms for clustering data stream have been systematically studied and comprehensively summarized. On the basic of previous research, the novel algorithms for fractal clustering data stream are proposed.Firstly, the thesis introduces some basic knowledge of data mining and fractal and clustering and some classical algorithms for clustering data stream.Secondly, the thesis present an algorithm which is based on fractal to cluster data stream and uses the change of fractal dimension to measure the self-similarity between data and clusters .With noisy condition, the algorithm can discover arbitrary shape clusters that reflect the natural group status of data stream.The experiments show the good performance and effectivity of FClustream.Finally, the thesis conduct a comprehensive summary and discussion of Fclustream. The main problems of Fclustream and future research directions are discussed.