CNAK: Cluster number assisted K-means

Determining the number of clusters present in a dataset is an important problem cluster analysis. Conventional clustering techniques generally assume this parameter to be provided up front. %user supplied. %Recently, robustness any given algorithm analyzed measure stability/instability which turn determines number. In paper, we propose method analyzes stability for predicting Under same computational framework, technique also finds representatives clusters. The apt handling big data, as design using \emph{Monte-Carlo} simulation. Also, explore few pertinent issues found clustering. Experiments reveal that proposed capable identifying single cluster. It robust high dimensional and performs reasonably well over datasets having imbalance. Moreover, it can indicate hierarchy, if present. Overall have observed significant improvement speed quality numbers composition large dataset.