Improving geocoding outcomes for the Nebraska Cancer Registry: learning from proven practices.

This report summarizes geocoding improvement experiments in the Nebraska Cancer Registry. An initial assessment of previous geocoding suggests that some proven geocoding procedures have not been followed, and overall results were unacceptable. This study concluded that when updating different address files from different time periods, it is sufficient to use the most recent street centerline database. The combination of match score of 80 and spelling sensitivity of 80 in ESRI's ArcGIS geocoder is sufficient for most geocoding purposes. Given the sizable number of unmatched addresses, the Google Maps geocoding service was used. A comparison of 1500 high-quality addresses that were matched by both Google Maps and ArcGIS geocoders shows that, in most cases, the location discrepancies between the two were acceptable. The median distance between each pair of 1500 coded locations was 36.6 meters, with an average of 92.8 meters. Distance discrepancies were larger in urban fringe areas and smaller toward urban centers. It was concluded that by strictly following proven procedures including address coding specification, Internet-based White Pages for reverse address finding, and Internet-based geocoding, a 90% or even a 95% match rate is achievable.