Prediction of 'artificial' urban archetypes at the pedestrian-scale through a synthesis of domain expertise with machine learning methods

TL;DR

This paper combines urban planning expertise with machine learning to classify towns as 'artificial' or historical, providing a new tool for assessing urban development at the pedestrian scale.

Contribution

It introduces a novel synthesis of domain knowledge and machine learning to create classifiers that distinguish between artificial and historical towns using pedestrian-scale data.

Findings

Supervised classifier identified 185 artificial towns.

Models align with urbanist intuitions.

Potential for improved urban development assessment.

Abstract

The vitality of urban spaces has been steadily undermined by the pervasive adoption of car-centric forms of urban development as characterised by lower densities, street networks offering poor connectivity for pedestrians, and a lack of accessible land-uses; yet, even if these issues have been clearly framed for some time, the problem persists in new forms of planning. It is here posited that a synthesis of domain knowledge and machine learning methods allows for the creation of robust toolsets against which newly proposed developments can be benchmarked in a more rigorous manner in the interest of greater accountability and better-evidenced decision-making. A worked example develops a sequence of machine learning models that distinguishing `artificial' towns from their more walkable and mixed-use `historical' equivalents. The dataset is developed from network centrality, mixed-use, land-use accessibility, and population density measures as proxies for spatial complexity, which are computed at the pedestrian-scale for 931 towns and cities in Great Britain. Using officially designated `New Towns' as a departure point, a series of clues is then developed. First, using an iterative human-in-the-loop procedure, a supervised classifier (Extra-Trees) is cultivated from which 185 `artificial' locations are identified based on data aggregated to respective town or city boundaries. This information is then used to train supervised and semi-supervised (M2) deep neural network classifiers against the higher resolution dataset. The models broadly align with intuitions expressed by urbanists and show potential for continued development to broach ensuing challenges pertaining to: selection of curated training exemplars; further development of techniques to accentuate localised scales of analysis; and methods for the calibration of model probabilities to align with the intuitions of domain experts.