To group at the base of users´ usage preference of network services based on fast hierarchical clustering algorithm

Because possessing the huge superiority on selective services marketing ^[1], which could bring tremendous research value, user´s usage preference of network services is becoming an issue. But there was seldom research on grouping services. In this paper we investigate the grouping services, and concretely study the clustering algorithm, which based on the users´ usage preference of network services grouping, and compare the time complexity and the clustering results of classical clustering algorithms, and choose the hierarchical clustering algorithm to group the network users according to the characteristics of analytical data and the analysis of demand. Meanwhile, as to the high time complexity of classical hierarchical clustering algorithm, we improved it by introducing a fast hierarchical clustering algorithm, which could merge many data samples at a time based on entropy grouping and data characteristics, and this algorithm significantly reduce the time complexity. Research results provide a specific grouping for services preference. In this way, data is provided for selective management and commercial package customization.