Influence Analysis of Linear Data Distribution on Different Clustering Results
Yu Chen Song, Xiao Liang Jia, Hai-Dong Meng · Advanced materials research · 2012
Through simulating kinds of practical data distribution forms, we could construct different linear data distribution based on three-dimensional space. According to the widely applicable clustering methods based on traditional methods classification, we could compare and analysis of different clustering results on kinds of linear distribution data. Firstly, we could conduct the clustering experiments of simple linear distribution data; secondly, according to the experience above, we could conduct the clustering experiments of complex linear distribution data. At the same time, combining mature computer visualization technology of displaying clustering results with the widely applicable evaluation index of interpretative statement, we could come to the conclusion that relevant influences of linear data distribution on clustering results of different clustering methods, proposing more optimized clustering analysis process on different linear data distributions.