Proportional data clustering using K-means algorithm: A comparison of different distances
Jai Puneet Singh, Nizar Bouguila · 2017
In this paper, we discuss proportional data clustering. It emerges In many applications such as document clustering and Image classification using bag of visual words approach. When deploying mixture models for clustering, there Is always a problem of initialization, and It Is common to initialize using K-means algorithm. In proposed work, we present K-means clustering approach using different distance metrics. In particular, we propose the consideration of the Altchlson distance. Experimental results are presented using silhouette plots for showing divergence from the center, and confusion matrix Is used to validate our clustering of synthetic and real data sets of Images and texts. The algorithm with Altchlson distance metric results Into lower error rates.