An Incremental Shared Nearest Neighbour Clustering Approach For Numerical Data Using An Efficient Distance Measure

Abstract

Clustering is one of the prominent fields of data mining. A major drawback of traditional clustering algorithms is that they perform clustering on static databases. But in real time databases are dynamic. Therefore incremental clustering algorithms have become an interesting area of research wherein clustering is performed on the incremental data without having to cluster the entire data from scrape. In this paper, a new incremental clustering algorithm called Incremental Shared Nearest Neighbor Clustering Approach (ISNNCA) for numeric data has been proposed. This algorithm performs clustering based on a similarity measure which is obtained from the number of nearest neighbors that two points share. In order to identify nearest neighbors, a distance measure is used. A distance measure that performs well with this algorithm has been identified in this work. This algorithm could find clusters of different shapes, sizes and densities.