The Level of K-means Clustering Algorithm Based on the Minimum Spanning Tree

Aiming at the number of clusters and the problem of the clustering results instability result of random selection of the initial cluster centers.Combine minimum spanning tree(MST) with the thinking of split and cohesion of hierarchical algorithm, proposed a MST-based hierarchical K-means algorithm. A MST is generated by sample data at the time of the algorithm initialization, and then use the thought of disunion, divide the data into smaller cluster. Gain the evaluation function value of each operation by iterative operation of the K-means algorithm, determine whether to cluster merging with value of evaluation function, to futher determine the clustering number of cluster.Experimental results show that the algorithm can more accurately determine the number of clusters, and the stability of the clustering results of the algorithm is better than the basic K-means.