The Shared Nearest Neighbor Algorithm with Enclosures (SNNAE)

Unsupervised learning is that part of machine learning whose purpose is to find some hidden structure within data. Typical task in unsupervised learning include the discovery of “natural” clusters present in the data, known as clustering. The SNN clustering algorithm is one of the most efficient clustering algorithms which can handle most of the issues related to clustering, like, it can generate clusters of different sizes, shapes and densities.This paper is about handling large dataset, which is not possible with existing traditional clustering algorithms. In this paper we have tried an innovative approach for clustering which would be more efficient or rather an enhancement to the SNN (Shared Nearest Neighbor) and we are going to call it ‘Shared Nearest Neighbor Algorithm with Enclosures (SNNAE)’. The proposed algorithm uses the concept of ‘enclosures’ which divides data into overlapping subsets and provides a better output than the SNN algorithm. The experimental result shows that SNNAE is more scalable, efficient and requires less computation complexity compared to SNN.