Dynamic Parameters Of A Density-based Unit Clustering Algorithm

In this article, concepts, techniques and algorithms about clustering will be discussed. We give a dynamic parameter solution for parameter selection problem in density based clustering algorithm.Among the various algorithms put forward, a main class of them are based on "distance" , whether it is in the sense of traditional Eculid distance or others. "K-means" and "k-medoids" are two of this kind. However, these algorithms are inefficient when dealing with large data sets and data sets of high dimension. Further more, the number of clusters they can find usually depends on users' input. But this task is often a very tough one for the user.In this article we give a solution for this parameter selection problem,called a dynamic parameter computing algorithm. The algorithm in this article differs much with above ones and it takes a totally different approach, which we call a grid and density based algorithm. It can automatically find out subspaces containing interesting patterns we want and discover all clusters in that subspace. Besides, it performs well when dealing with high dimensional data and has good scalability when the size of the data sets increases. As results, clusters found are presented to users in DNF expressions.