HCIL-2008-26

Analyzing multivariate geospatial data is a non-trivial task. Geographical location is a variable in itself that might be correlated with other fields in the data. When looking at the variable distribution in a geographical region, users often need to identify areas where the variable is overrepresented. We call such local maxima ”hotspots” and present Fervor: a hotspot explorer. Fervor visualizes hotspots as a continuous heat map and offers a range of tools to investigate relationships within the underlying data. We adopted and expanded the rank-by-feature framework [12] to quantify the strength of relationships between the different fields describing the data. In this paper, we use Fervor to investigate traffic incident data in the state of Maryland.