Locating household profiles in a polycentric region to refine the inputs to an agent-based model of residential mobility

In this paper we present a method to analyse data from observed population statistics in order to create a stylised location typology based on household profiles (micro-geographic level). The household profiles are used to inform the development of rules for an agent-based model (ABM) of residential population mobility. The method is demonstrated with the 2001 population census data for East Anglia (EA), UK. Household profiles are derived from a principal components analysis to reduce dimensionality in the census data combined with a cluster analysis to aggregate the observations. We test whether a range of household archetypes that link specific locations to specific profiles can be identified simultaneously at both meso- and macro-geographic levels. Results reaffirm conventional archetypes: single people concentrate in city centres, families mainly locate in the suburbs and peri-urban areas and retirees are more prevalent in the suburbs and close to the coast. Surprisingly, however, there is a clear distinction between locations with cohabiting couples (located at the fringe of the periurban areas) and locations with married couples (located in the countryside). The observed spatial distributions reveal different location choices between different household profiles. Identification of these household profiles supports the development of residential mobility models that may be used to explore regional and local planning issues with respect to future population projections and environmental change.