Geo-coding addresses in historic British census data: An open methodology

Abstract

This article introduces CensusGeocoder, an open-source python package for automated address geo-coding, and AddressGB, an openly available dataset of c. 121 million individuals with geo-coded addresses from I-CeM (digitized historic British census data for 1851 to 1911). AddressGB enables address and street-level GIS analysis of I-CeM data, a much higher spatial resolution than existing approaches that aggregate individuals to parishes or census registration sub-districts. This also opens new opportunities for linking historic census data to other spatial datasets. The article details the methodology underpinning CensusGeocoder and showcases the scale, accuracy, representativeness, and future applications of AddressGB.