**Abstract:**
Recently, so many methods have been invented to determine issues with the choice of starting points in K-Means clustering algorithm. The Global K-Means and the Fast Global K-Means algorithms both are basis of such methods. They frequently insert one cluster centre at a time. The Weighted Fuzzy C-Means algorithm is as well extremely admired for fuzzy basis data clustering. However these all clustering methods are immensely influenced through the extreme environment of high dimensional data values. Every data in the dataset has compound characteristics and the cost of some characteristics might be so huge that the significance of additional characteristic costs might be entirely overlooked in the clustering procedure. The complexity of utilizing high dimensional datasets in clustering process is well known. To resolve these difficulties and to get better clustering algorithm for huge high dimensional datasets we proposed an algorithm “an enhanced global k-means (EGKM) algorithm for cluster analysis”. To calculate the performance of the both FGKM and EGKM algorithms we use six datasets: Letter, Car, Iris, Kddcup, Nursery, Ozone and Spambase. Our trial study shows that EGKM performs much better than FGKM for every high dimensional datasets.

**Keywords:**
GKM, FGKM, K Means, Cluster Analysis.