An Unsupervised Natural Clustering with Optimal Conceptual Affinity
Gary C. Barker · Journal of Intelligent Systems · 2010
One Unsupervised Natural Clustering Algorithm with Optimal Conceptual Affinity has been designed and developed.Unlike the conventional Optimal Clustering Algorithm and its several versions, which require the Conceptual Affinity or threshold distance from the user as input, this algorithm performs Natural Clustering of the input data set by automatically finding out the Conceptual Affinity or threshold.The clustering is completely Unsupervised as the users do not even have to put the Conceptual Affinity for Natural Clustering-the system on its own will ultimately find it.Thus the system uses that computed Conceptual Affinity or threshold and exhibits the natural clusters as output.The system contrasts heavily from the conventional clustering techniques, which require some parameters for clustering and are not guaranteed to form natural number of clusters.