Symbolizing point data in SpatialKey
2 September 2008
SpatialKey is especially well-suited at representing point datasets with thousands, even tens of thousands, of rows. Symbolizing such datasets offers many cartographic challenges. Rendering the individual points is the simplest strategy but quickly leads to lost data via overlap and cognitive overload due to the sheer number of displayed points. Further, symbolizing points with points only tells the user one thing about them: where they are. Many datasets include attributes (sale price of a home, age of a cancer patient, number of prior offenses of a suspect) that can be symbolized as well to aid in geographic analysis.
The standard cartographic approach to symbolizing point data with attributes attached is the proportional symbol map. Such maps, which are one of the three symbolization methods currently offered in SpatialKey, use symbols (typically circles) scaled proportional to point values.
Offering multiple symbolization options — each available at all times and map scales — allows users to switch to a more appropriate rendering for their dataset and switch back-and-forth to see their data in new ways. In addition to the proportional symbols shown above, our templates currently offer two raster-based aggregated renderings: the heat grid and heat map.
In the above screenshots, the colored heat grid and heat map represent the density of Wal-Mart stores. In this case, store locations are aggregated to an arbitrary, scale-dependent grid. Thus, the brightest grid squares (and the hottest areas on the continuous heatmap representation) represent the areas of highest Wal-Mart concentration.
These renderings can also be used to show attributes of point data, in which case the hottest/brightest areas represent the areas with the highest average or total value for a given attribute. And of course mousing over the map reveals tooltips with exact values for these “quadrats” (screenshot below shows Sacramento home sales data).
The heatmap symbolization is newer, less vetted by the cartographic community, and perhaps less straightforward than the proportional symbol and heat grid renderings. But with the proper user interface, and in concert with other available renderings, we believe heatmaps can help facilitate effective visualization of geotemporal data.
We are particularly interested in feedback on the use of the heatmap symbology for mapping attributes of data (like home prices). We are excited to add additional symbolization options to our templates, as well as improvements to existing renderings, as we continue to develop the SpatialKey visualization system.
1 Comment