Ksummary Analysis Method Based On Adaptive Multiple Clustering

Data mining is a process, it can find implicit, novel, the decision-making process that has a potential value of knowledge and rules in database. It has been widely used in many fields. Clustering analysis is one of the most important technology in data mining field and it has achieved fruitful research results on the theory and method.Wireless network data is an important part of telecom data. Mainly is information collected by the base station, contains attributes such as the user id and call quality. By research and analyze the data,we can not only get the user’s information, such as user demand for communication service, but also help enterprises after knowing the user communication quality, so provide better services for user. Wireless network this huge data set, analysis of this data set is the use of data mining field, and can divide it into several clusters by appropriate clustering. According to the characteristics of clustering, the similarity between the clusters is higher, the degree of similarity between the cluster is low. analysis and research could be done according to the specific information within the clusters, so as to get useful knowledge. According to the characteristics of the data set that contains the numerical classification and attribute of mixed attribute set, select the appropriate clustering algorithm for clustering result is of crucial importance.There has been several specific kinds of clustering algorithms.each kind of clustering algorithm has its own scope of application, hat is to say, different clustering is suitable for different data sets. KSummary algorithm is a kind of clustering algorithm, and can better deal with classification attribute and mixed attribute data. The algorithm is put forward with the information represents a cluster center, high frequency value compared with the attribute value to represent the method of cluster, deviation smaller, especially in the case of different values of frequency difference. But the algorithm also has some disadvantages: cluster number K need given in advance; The algorithm is sensitive to initial value; The algorithm may be trapped in local optimal solution.KSummary algorithm has two shortcomings when the data quantity reaches a certain level, failing in defining the number of clusters and the initial centers of clusters. An algorithm based on adaptive multiple clustering is proposed in this paper. Hierarchical clustering algorithm is used to define the number of clusters and density clustering algorithm defines initial centers of clusters. At last, KSummary algorithm is applied to produce the final clustering results. Experimental results and theoretical analysis show that our method is more efficient than traditional method.Finally, the adaptive multi-trip cluster analysis method is applied to the telecommunications industry, wireless network data sets. Factors affecting the quality of communication are many, in order to more accurately analyze the importance of the information in the uplink and downlink of these factors, the two separate pins for the two data sets, respectively, of the two sets of data clustering. Analysis of two clusters of clusters of information, cross-comparison of the two clustering results in clusters with similar characteristics, obtained experimental analysis charts, analysis of the reasons, and enterprise can set the appropriate station layout for their equipment according to these information.