An Efficient Dimension Reduction Technique for Basic K-Means Clustering Algorithm

K-means clustering is being widely studied problem in a variety of applicationdomains. The computational complexity of the basic k-means is very high, the number ofdistance calculations also increases with the increase of the dimensionality of the data.Several algorithms have been proposed to improve the performance of the basic k-means.Here we investigate the behavior of the basic k-means clustering algorithm and twoalternatives to it, we have analyzed the performances of three different standardizationmethods. Equivalently, we prove that z-score and principal components are the best preprocessingmethods that will simplify the analysis and visualize the multidimensionaldataset. The analyzed result revealed that the z-score outperform min-max and decimalscaling also principal component analysis picks up the dimensions with the largestvariances. Our results also provide effective ways to solve the k-means clusteringproblems.