Experimental Research on Impacts of Dimensionality on Clustering Algorithms

Experiments are carried out on datasets with different dimensions selected from UCI datasets by using two classical clustering algorithms. The results of the experiments indicate that when the dimensionality of the real dataset is less than or equal to 30, the clustering algorithms based on distance are effective. For high-dimensional datasets--dimensionality is greater than 30, the clustering algorithms are of weaknesses, even if we use dimension reduction methods, such as Principal Component Analysis (PCA).