A multifaceted comparison of ArcGIS and MapMarker for automated geocoding

Area of Science:

• Public Health
• Spatial Epidemiology
• Geographic Information Systems (GIS)

Background:

• Geocoding is crucial for public health surveillance and spatial epidemiology in the USA.
• Public health departments utilize geocoding for disease outbreak investigations and assigning health records to geographic units.

Purpose of the Study:

• To evaluate and compare the performance of ArcGIS and MapMarker for automated geocoding of residential addresses from health administrative data.
• To understand the features, performance, and limitations of these two geocoding software packages.

Main Methods:

• Comparison of ArcGIS and MapMarker using automated geocoding of 551,798 residential addresses from New York State health data.
• Evaluation based on completeness (match rate), geocode similarity, and positional accuracy.

Main Results:

• ArcGIS achieved a higher match rate (76.3%) compared to MapMarker (57.7%).
• High similarity between geocodes from both packages was observed, particularly in urban and suburban areas.
• ArcGIS demonstrated superior positional accuracy when geocodes differed by over 100m, with its points closer to parcel centroids.

Conclusions:

• ArcGIS, particularly its composite address locator, offers better geocoding results by utilizing reference data more effectively.
• While positional differences were minimal, combining results from both geocoding methods can maximize match rates and reduce manual effort.