Performance evaluation of K-means clustering algorithm with various distance metrics

Abstract

Data Mining is the technique used to visualize and scrutinize the data and drive some useful information from that data so that information can be used to perform any useful work. So clustering is the one of the technique that has been proposed to be used in the area of data mining The notion behind clustering is to assigning objects to cluster based upon some customary characteristics such that object belonging to one cluster are similar other than those belonging to other clusters. There are numerous clustering algorithms available but K-means clustering is widely used to form clusters of colossal dataset. The footprint factor for k-means clustering is its scalability, efficiency, simplicity. This proposed paper aims to study the k-means clustering and various distance function used in k-means clustering such as Euclidean distance function and Manhattan distance function. Experiment and results are shown to observe the effect of these distance function upon k-means clustering. The distance functions are compared using number of iterations, within sum squared errors and time taken to build the full model.