DDG-clustering: a novel technique for highly accurate results
Zahraa Said Ammar, Mohamed Medhat Gaber · 2009
A key to the success of any clustering algorithm is the similarity measure applied. The similarity among different instances is defined according to a particular criterion. State-of-the-art clustering techniques have used distance, density and gravity measures. Some have used a combination of two. Distance, Density and Gravity clustering algorithm “ DDGClustering ” is our novel clustering technique based on the integration of three different similarity measures. The basic principle is to combine distance, density and gravitational perspectives for clustering purpose. Experimental results illustrate that the proposed method is very efficient for data clustering with acceptable running time.