Improved K-Means Clustering Algorithm by Getting Initial Cenroids
Ghousia Usman, Usman Ahmad, Mudassar Ahmad · 2013
2 Abstract: To extract useful information from huge data sets are considerable problem for researcher that is insufficient for conventional databases querying methods. K-means clustering algorithm is a one of the major cluster analysis method that is commonly used in practical applications for extracting useful information in terms of grouping data. But the standard K-means algorithm is computationally expensive by getting centroids that provide the quality of the clusters in results. For improving the performance of the K-means clustering algorithm, several methods have been proposed that are described in literature review. This paper proposes a method for effective clustering by selecting initial centroids. Firstly, this algorithm evaluate the distance between data points according to criteria; then try to find out nearest data points which are similar; then finally select actual centroids and formulate better clusters. According to the results of new solution, the improved k-means clustering algorithm provides more accuracy and effectiveness rather than previous one.