A Novel Fast Clustering Algorithm

SNN is a shared nearest neighbor based clustering algorithm. It is improved to process the data with categorical attributes and be given a simple and definite method to select threshold of the algorithm. By combine one-pass clustering algorithm with the enhanced SNN clustering algorithm, we present a fast clustering algorithm which can find different sizes, shapes and densities in noisy, high dimensional and large dataset. The time complexity of the presented clustering algorithm is nearly linear with the size of dataset. The experimental results on real datasets and synthetic datasets show that the clustering algorithm is effective, robust and practicable.