Statistical Examination of Density-Based Spatial Clustering for Data with Multiple Dimensions

Abstract

Spatial data mining is a method for extracting useful information from geographical databases. Collecting massive volumes of geographical data for uses as diverse as geospatial analysis and biomedical research has made this a very demanding area of study. There has been an exponential growth in the quantity of geographic data collected. Therefore, it was incomprehensible to humans. In the realm of geographic databases, clustering has emerged as a key data mining technique for discovering new information. There has been a lot of focus on, and proposal of, novel clustering algorithms for development in recent years. An early density-based clustering method was DBSCAN. It is capable of extracting clusters of varying sizes and forms from massive datasets that include noise and outliers. Using synthetic two-dimensional geographical data sets, this study presents the findings of assessing the qualities of density-based clustering characteristics of three clustering algorithms: DBSCAN, k-means, and SOM.