New optimized agglomerative clustering algorithm using multilevel threshold for finding optimum number of clusters on large data set

The traditional hierarchical clustering methodswhich are suitable to categorize the large dataset but do not produce effective partitions result, because of their high computational complexity, higher number of iterations and more misclassification errors. To overcome this, in this paper, a new technique called Optimized Agglomerative Clustering(OAC)is proposed for effective clustering and pattern classification of large data set, and we also introduced three new threshold methods that can find optimal merge cost for limit the number of iterations. The proposed OAC technique is work with three new thresholds that aims to automatically produce optimum number of clusters with good accuracy, and it reduces computational complexity, numbers of iterations and misclassification errors. Experimental results show that the proposed technique OAC produces a better clustering result with high accuracy than traditional techniques.