Research On Clustering Algorithm Based On Fish Group And Rough Set

21st century is the era of information and data, with the rapid growth of computerization of society a lot of data pouring from every corner of our daily-life. The growth of data in different sectors rendering a geometric growth that is faster than ever before, which not only improves the people's interest of data processing technology and also leads to the discipline of data mining drawn more and more attention of the society. Cluster analysis as an important part of data mining technology, which has been rooted in many application areas such as biology, security, business intelligence and Web searches.This article discusses the artificial fish-swarm algorithm, K-medoids algorithm,granular computing, rough sets theory and main research jobs are listed as follows:Fish-swarm algorithm and K-medoids algorithm are analyzed advantages and disadvantages, a hybrid algorithm based on artificial fish-swarm and K-medoids was proposed. Firstly, an artificial fish-swarm algorithm was improved. Secondly, K-medoids was combined with improved artificial fish-swarm algorithm for clustering. The experiment shows that the algorithm is better than standard fish-swarm algorithm and traditional K-medoids algorithm in accuracy and stability on clustering effect.The effect of rough set and granular computing to clustering algorithm is analyzed, an artificial fish-swarm algorithm based on rough sets and granular computing was proposed.Initially, the algorithm introduced the granular computing theory and initialized the fish group by the density and max-min distance means so that the algorithm was effected by random.Meantime the algorithm combined with rough set and attribute reduction and decision system to resolve the clustering problem of boundary data. As the termination condition of the algorithm the fitness function designed of the class within the compactness and inter-class separability principle. As experiment results show that the algorithm has enhanced the accuracy and the abilities to obtain global extremum and embodied better clustering performance.