Mixed Clustering Analysis Based On Improved Particle Swarm Optimization And Improved K-harmonic Means

With the advent of the big data age,people put high demand for dealing with massive,heterogeneous high-dimensional data,the researchers hope to dig out useful knowledge more intelligently and quickly.The clustering analysis is an important data mining method,and the clustering analysis algorithm is studied and optimized to improve the clustering effect,which has high value in practical meaning.In this paper,the writer focuses on the fusion clustering of intelligent algorithms and clustering algorithms,then improves the Particle Swarm Optimization algorithm(IPSO),and optimizes the parameters of K-harmonic means(KHM),This paper studies the fusion of algorithms and proposes a fusion clustering algorithm IPSO-KHM.Specific research work can be divided into the following sections:Firstly,analyzing the principle and parameter setting of the clustering algorithm KHM in detail.In order to solve the problem of poor clustering effect in high dimensional data scene,the parameter setting is discussed and optimized.Secondly,the intelligent algorithm is analyzed deeply,the locally convergent particle shape of the PSO algorithm is analyzed.obtaining Gaussian distribution of the particles.The mutation concept of the GA algorithm is introduced to improve the local optimal performance of the particles.Combining the particle space density Method,proposed a mutation strategy related to the search efficiency of the particle,and generates an IPSO algorithm to optimize its global search capability.Thirdly,in order to improve the performance of clustering,convergence speed and classification accuracy,the writer proposed a fusion algorithm IPSO-KHM and designed algorithm flow.Fourthly,the standard UCI data set is used to analyze the IPSO-KHM,and the feasibility and validity of the algorithm are verified.The results of this paper have better comprehensive performance,the performance of clustering effect,convergence speed and classification accuracy are better than other algorithms.