Research On Adaptive Elastic Network Algorithm For Cluster Analys Is

With the coming of large data age,the requirements of the cluster analysis tools are getting higher and higher.The existing clustering algorithms include hierarchical clustering algorithm,clustering algorithm,density-based clustering algorithm,grid-based clustering algorithm and model-based clustering algorithm,but the applications of these algorithms has great limitations and is difficult to meet the needs of existing data clustering analysis.In order to meet the needs of high dimensional large data sets in clustering analysis,this thesis presents a new clustering algorithm.The algorithm can overcome the shortcomings of the traditional clustering algorithm and reduce the influence of the noise data on the clustering results.At the same time,the algorithm can be applied to the clustering problem of large data sets.This thesis first introduces the concept and background of data mining and clustering analysis,and describes the importance of clustering algorithm in data mining and its wide application in various fields of science.Secondly,some typical clustering algorithms are introduced in detail,the idea of the algorithm is described,and the advantages and disadvantages of the algorithm and the applicable data types are analyzed.Then,this thesis introduces the elastic network algorithm in deta il,analyzes the algorithm's drawbacks,and introduces an improved elastic network algorithm with time-varying parameters.Next,this thesis presents the Elastic+K-means algorithm and AEN algorithm.Finally,in the thesis,the K-means algorithm,the K-medoids algorithm,the Elastic+K-means algorithm and the AEN algorithm are separately used to cluster the typical data set,2D,3D,4D,7D,10 D and 13 D random data sets and standard data sets in order to compare the algorithms performance.At the same time,the AEN algorithm is tested in low dimension data sets and high dimension data sets to analysis the quality of the algorithm.