Ironing out the kinks in angular network analysis

New research published on theoretical network analysis

Spatially localized closeness and betweenness are the bread and butter of modern spatial network analysis, and in particular, spatially localized angular analysis has become something of a de facto standard.  But did you know this standard contains a fundamental contradiction?

Cooper, C. H. V. (2015) Spatial localisation of closeness and betweenness measures: a self-contradictory but useful form of network analysis.  International Journal of GIS (accepted, in press).

This paper arose out of an issue we discovered during the design of sDNA.  Spatial localization usually implies restricting analysis to a certain (Euclidean) scale, but angular shortest paths are by definition no shorter than Euclidean ones, and are longer at least some of the time.  Thus angular shortest paths break the specification of spatial locality.

Does this matter?  Usually no, but sometimes yes.  In particular, weighted angular analysis can suffer from some serious errors if the wrong algorithm is used to handle the contradiction.  sDNA+ allows a choice of algorithms for just this purpose.