Analysis And Optimization Of Fuzzy Clustering Algorithm Based On Data Stream

In recent years,with the continuous development of science and technology,a kind of data which is different from the traditional concept has been produced.These data are dynamic,high speed,massive,real-time and continuous.We call this new type of data ‘data flow'.Data stream has played an important role in many filed,such as wireless senor network,network intrusion financial analysis market and so on.As a very effective tool for data analysis,cluster analysis has become a hot spot in the research field of data mining,through clustering analysis of the data,we could find that there is valuable information in the mass data.At present,many clustering algorithms are aimed at the processing of traditional data sets,however,due to the characteristics of the data stream,traditional clustering algorithms show a lot of deficiencies in the data stream processing.Therefore,it is necessary to design a new clustering method,which can better handle the data stream.In this paper,we first introduce the fuzzy c-mean clustering algorithm,but this algorithm is a local search algorithm,if the initial value is not appropriate,a local optimal solution is obtained finally.Aiming at the deficiency of FCM algorithm,we propose a new algorithm called GSA_FCM.The algorithm can be applied to clustering data stream,which use the weight decay to reduce the influence of the historical data for the clustering results,at the same time,by combining the genetic simulated annealing and fuzzy c-mean clustering algorithm,the new algorithm can quickly converge to the global optimal solution and overcome the sensitivity to the initial clustering center,and through the experiments to show that the clustering quality of GSA-FCM algorithm is more reliable.Finally,the local spatial information of data points is considered in this paper,proposing a Multi-objective simulated annealing based fuzzy-clustering for data stream clustering.The algorithm can also be used for clustering data stream,and effectively reduce the effect of noise points on clustering quality,while the cluster edge can be more accurate,and finally compare the performance of algorithms through experiments.