Geographical Visualization of Rare Events

Maps contain and effectively visualize a number of spatial information. Advances in GIS enable researchers to analyze and represent spatial information through digital maps. Choropleth maps represent different quantities showing usually rates, percentages or densities. Generally, researchers make choropleth maps using raw rates. But, if the events are rare, raw rates cannot be sufficient in representing spatial phenomena. That is to say, if the population is large and events are rare, we cannot be sure that the raw rate is correct. The objective of this study is to make choropleth maps by several rate calculation methods and compare them. We use three methods in choropleth mapping; a raw rate, empirical Bayesian method, and spatial rate method which use prior probabilities. The experiments reveal that maps are somewhat different by used methods. We suggest that a raw rate method can not be an only way to make a rate map and researchers should choose an appropriate method for their objectives.