Concentric Cluster Analysis: A Radial Approach for Quantifying Clustering and Scatter in Two-Dimensional Data

Assessing the spatial clustering or scatter of points in two-dimensional data is essential for applications across scientific, engineering, and multimedia domains. Conventional metrics, such as variance, convex hull area, and pairwise distances, provide global measures of spatial dispersion but may overlook local or radial structural nuances. This paper introduces Concentric Cluster Analysis (CCA), a novel approach that segments data into concentric rings around the centroid to quantify radial distribution patterns. The methodology is validated on both synthetic datasets and a real-world shrimp feed tray image, representing practical aquaculture monitoring. CCA generates interpretable sub-metrics from both histogram-based and cumulative-based analyses, offering granular insights into the degree and nature of clustering. Comparative experiments show that CCA complements and extends conventional spatial metrics, effectively distinguishing between compact and dispersed point patterns and revealing subtle features that may be masked by aggregate measures. The results establish CCA as a valuable tool for spatial data analysis, with demonstrated applicability to real-world multimedia image analysis and potential for broader use in pattern recognition, spatial clustering, and environmental monitoring.