Abstract:

Several K-means algorithm are available for clustering using various datasets of simulation. Yeast dataset and iris dataset are used for clustering by using K-means algorithm with numerous iteration and lower accuracy. For clustering, K-means algorithm of these datasets are simulated by enhanced version, Minimum spanning tree approach is used by Improved K-means algorithm. Each and every input data points are produced by an undirected graph and later shortest distance is computed by intern outcomes with increased accuracy with lower number of iterations. Java programming language was used for the simulation by these algorithms. Analysis and comparison of both the algorithms are resulted. Algorithms are run many times below various groups of clusters. From the outcomes, it is observed that better performance are acquired by improved K-means algorithm and contrasted with L-means algorithm. Accuracy is increased by the values of number of clusters. It is inferred that k’s specific value increases the accuracy of this algorithm with optimal values.

Keywords: