ClusterLens: A focus+context approach to multi-resolution spatial clustering

Spatial clustering can reduce visual clutter on maps and facilitate pattern recognition. However, interactive map exploration needs the spatial clustering to be a dynamic generation and representation process. Users may change derivative representations of clusters during the exploration process. To address this issue, we present ClusterLens, a new interaction technique that brings a focus+context approach into multi-resolution spatial clustering process. A lens is laid on a base map to avoid occlusion with the original and actual point locations. The lens can aggregate the data points at various spatial resolutions as map zoom level changes. We propose three primitives of resolution for spatial clustering: heatmap, circle, and grid, to generate and represent clusters in a separate mapping system. The lens and the base map are linked at all times. We also incorporate coordinated views into the ClusterLens system to facilitate context switching and comparison. We discuss the applicability of our technique and present a use case where ClusterLens can be useful to explore data distributions and reveal spatial patterns.