Clustering Analysis Of Big Data Based On The Minimum Spanning Tree Of Network Optimization

With the high-speed development of the social, science and Internet technology, more and more data brithes from all aspects with an exponential growth, the big data of high dimensional and complex have occupied all of our society. It is necessary for us to dig out the implied value behind these big data efficiently and accurately. Clustering analysis is one of the core technology of data mining algorithm. In addition, clustering analysis also be used as a preprocessing step of data mining in initial phase. However the traditional clustering algorithm exists many shortcomings and defects.Clustering analysis of big data based on the MST of network optimization rely on the graph theory, through analysising the relationship between each adjacency matrix structure of data objects to constructs a graph and to generate a MST of the graph. According to the actual problem and the distribution of these data, we cutting the edge of the MST according to the size of the length from big to small in turn. So, we obtain the sub tree of the MST, every sub tree is a optimal clustering. In this paper, we test these clustering algorithms with the international general UIC test data of IRIS, and the simulation experiments show that the MST algorithm based on network optimization is advantages than others.Finally, some problems and developments of big data mining are discussed.