Cluster Analysis of Data with Reduced Dimensionality: An Empirical Study

摘要

Cluster analysis is an important high-level data mining procedure that can be used to identify meaningful groups of objects within large data sets. Various dimension reduction methods are used to reduce the complexity of data before further processing. The lower-dimensional projections of original data sets can be seen as simplified models of the original data. In this paper, several clustering algorithms are used to process low-dimensional projections of complex data sets and compared with each other. The properties and quality of clustering obtained by each method is evaluated and their suitability to process reduced data sets is assessed.