The ideal neighbourhoods of successful ageing: A machine learning approach

Ageing outcomes are shaped by not only health conditions, but also their interactions with the external environment. While the effects of some specific neighbourhood characteristics such as rurality on ageing have been evaluated in various studies, we still know little about the relative importance of particular natural and urban environments and how the impact varies at different stages of the ageing process. This article addresses these knowledge gaps by analysing survey data from 33 European countries using a machine learning method called multivariate regression trees (MRT). Multiple wellbeing indicators are combined to form an ageing profile for each individual in the survey. After studying these profiles using MRT, we find that generally the affordability of health facilities is a major determinant of life satisfaction, self-rated health condition and mental wellbeing for individuals in most age groups. Other important but age-specific determinants are neighbourhood safety and accessibility to cultural facilities and to green areas. In contrast, characteristics such as urbanity, transportation and air quality do not significantly influence ageing outcomes. Our findings lend support to the resources theory in explaining ageing outcomes and suggest that more resources may have to be directed to improve the affordability and quality of health care services, the policing services and the accessibility to cultural and green areas in order to achieve more favourable ageing outcomes.