A Novel Approach for Clustering Data Streams Using Granularity Technique

Data Stream mining has large scope due to their usage in vice variety of application and business purpose. It provides the meaning full usage information which use full to take decision and also for planning purpose. According to application needs on particular parameter consideration there will be change in clustering method use in a stream Data mining. The purpose behind survey paper is explore the widely use clustering method StreamKM++ beneficial over the different clustering method and resolve issues of traditional clustering. Also contain different clustering method like hierarchical, density base, Partitioning Method study, Parameter and their operational methodology. BIRCH is faster than StreamKM++ but output of it not efficient and same way compare it with StreamLS, which partitions input data stream into chunk and clustering each chunk base on local search. Outcome of that is quality comparable and StreamKM++ significant better scalable with number of cluster. Clustering method apply using 2-phase method. Setting the arrival rate of input stream Data using AIG, same way sets the memory for output using AOG, and setting processing to consume less resources using AIP. Using both method that's providing the better quality with respect to time clustering of stream data.