Frith, MJ; (2017) Using qualitative distance metrics in space syntax and configurational analyses. In: Heitor, T and Serra, M and Pinelo Silva, J and Bacharel, M and Cannas da Silva, L, (eds.) Proceedings of the 11th International Space Syntax Symposium (SSS11). Instituto Superior Técnico, Departamento de Engenharia Civil, Arquitetura e Georrecursos: Lisbon, Portugal.

Preview

Text Article SSS11.pdf - Published Version Download (694kB) | Preview

Abstract

One interesting result from space syntactic and graph theoretic configurational analyses are their ability to correlate with pedestrian and vehicular movement flows. These analyses function by representing the built environment as a graph of interconnected spaces. Network centrality measures, such as betweenness, closeness and choice, are then applied to quantify each space’s role in the network and, in this case, suggest the potential amount of movement. That said, because these three network measures are based on the calculation of shortest routes and attempt to model human movement, they should incorporate how we judge distances. In traditional space syntax distance is measured by the angular change accrued at junctions along a route and elsewhere it is often measured as the physical length of the route. However, and hitherto incorporated in these analyses, given the limitations of human cognition (e.g. Simon 1979), there must be some constraint to the cognitive precision of this spatial information. This idea of cognitive spatial imprecision is not new (e.g. Dutta 1988; see also Montello 2007) and a number of qualitative models have been produced that attempt to describe or emulate human spatial judgements. For example, Montello and Frank (1996) used simulations to test the ability of angular models to emulate real-life angle estimations, including an eight 45° cone model where angles are approximated to the nearest 45°. For qualitative physical distance the research is less developed however, for instance, Hernández et al. (1995) gives examples of metric distances being split into three, four or five categories. Based on this, the present study tests variations of qualitative angular and physical distance metrics by comparing their ability in network analyses to correlate with data on 409 pedestrian and 297 vehicular count observations in central London. The results indicate qualitative metrics can increase the correlation between network measures and movement flows. More specifically, angular qualitative metrics significantly improved a number of correlations between the network measures and pedestrian movement. For example, the eight 45° cone model improved the correlation for angular choice analyses of axial segments from 0.55 to 0.60. However, the tested qualitative metrics rarely or not at all (significantly) improved the correlations between angular analyses and vehicular movement and between metric network analyses and both types of movement. Reasons for these results are discussed with suggestions for future research.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item