Research Of K-means Clustering Algorithm Based On Fish Swarm

Data mining is a hot, which attracts lots of pursuers. Clustering analyzing is a function of Data mining, which uses clustering algorithm finding the clusters. Under the unexperiential knowledge, clustering algorithm converge different clusters based on the comparability of data. The result is the best comparability between the same clusters and the best variant between the different clusters.K-means algorithm is a popular partition method in cluster analysis the most widely used clustering criterion in. It is a method to find the center of two clusters between the unclassified data. The main merits are concision and celerity. When the cluster is denseness and has the obvious difference its result is the best effect. In disposing big dataset, K-means algorithm is relatively flexibility and efficiency. But people find that the search of the initial partition the selecting of the initial center will heavily affect the result of cluster to make it easy get hypodispel. The following question will be discussed: aiming at K-means algorithm that depends on initial value plunge local value not get global value, put forward ameliorate scheme. Introducing to one new swarm intelligence algorithm, Fish swarm algorithm, it only uses the aimed question function, have adapted ability of search space. Many fish collaterally search and advance efficiency. So many fish mend K-means algorithm that will make part local spots move to the global set.In the end, using experiment compare to the K-means algorithm and the mixed K-means algorithm. The result of experiment proves that Fish swarm algorithm overcome not only the question of K-means algorithm but also advancing the speed of executing.