DISC and K-Subspaces: Two Novel Algorithms to find Data Intrinsic Projected Clusters in Numeric Datasets

摘要：

Subspace data clustering is a promising research area of data mining. We are presenting two novel projected data clustering algorithms which compute subspaces on the basis of dimension intrinsic statistics. DISC algorithm computes one dimensional clusters and combine these one dimensional clusters on the basis of participating dimension weight in the computed subspaces to detect final clusters of data objects in the subspace. K-subspaces algorithm computes numerical analysis (DNA) of each dimension and use kmeans algorithm to cluster these DMAs. Each DNA cluster constitutes a subspace. Data objects of each subspace are clustered by using k-means algorithm. We tested proposed algorithms on real data sets. Both algorithms showed promising results. Here we are reporting results of two real world data sets for DISC algorithm and two results of clustering of real life data sets for k-subspace algorithm.

关键词： component Algorithm Data mining Subspaces Subspace data clustering

作者: Ariflqbal Umar Yunhong WANG Sadique Ahmed

作者单位: School of Computer Science & Engineering, BeiHang University Beijing China School of Computer Science & Engineering,BeiHang University Beijing China

会议类型: 国际会议

会议名称: 2010 International Conference on Software and Computing Technology(2010年软件与计算机技术国际会议 ICSCT 2010)

会议地点: 昆明

会议语种:英文

页码: 59-63

在线出版日期: 2010-10-17（万方平台首次上网日期，不代表论文的发表时间）