Hi (…)

I realise you are asking for detailed and technical information about the mechanics of how to run an angular analysis and how to define the “elements” of the model (axial lines, centre line…). I’m afraid I can’t help you on that because I don’t use a variety of representations or software tools.

However, I would like to mention something that perhaps you, or other people on this thread, may find of interest: the potential for “Node” analysis. Something that I think is worth bearing in mind when deciding on the representation technique.

Disclaimer: since I’m not actively involved in the Space Syntax community, other than occasionally reading these messages, it may be an issue already well analysed, in which case my apologies. But since I’m not sure, I thought I would share some thoughts, in case anyone sees potential for further research or to engage in conversation.

When I started looking into segments back in 1989, my main concern with the “traditional” (at the time) Axial Line representation was that streets, particularly long ones with plenty of connections, could show very different pedestrian flow patterns (or other social behaviour) along their length. Another concern was that the “weight" of each line was the same in the analysis regardless of their length or number of segments. Each line was an element with the same weight regardless of it being Oxford Street or a tiny side street.

On the other hand, I was very impressed with “axiality”, as a way of connecting elements that were not adjacent, and the strong results that integration measures provided when mapped against pedestrian flows.

To try to improve on that, and gain finer detail on the analysis, my suggestion was to use “nodes” and “segments” as the elements for the model and axiality as the relational property.

What I found was that, as a first pass, integration values for segments on the same line would be the same (segments are rather “dumb”, since they only “see” along the line) but integration values for nodes (the intersections between two or more lines) could vary quite significantly along a line. Basically, it was in the nodes where “stuff” happened, where connections took place. 

I then started to draw the axial lines in profile (with length along the X axis and integration value for the nodes along the Y axis), like a mountain chain. Drawn this way, one could see that the real integration value of the segments should relate to the nodes it had either side, (not necessarily those immediately adjacent, they could be tiny cul-de-sacs, but the highest one either side). I then derived an integration measure for the segments based on the profile of the line an the nodes either side, that improved correlation with pedestrian flows significantly.

Sorry for the long text, but the objective was to highlight that, in my view, the analysis of nodes, as spatial elements in their own right (and where the syntactic connections take place in the system), can provide a fine grain, and fairly rich, picture of the cities we want to analyse. 

I’ll be curious to know if anyone else out there has followed this line of work. 

Kind regards,
Juan Alayo

Note: The definition of “node" would need quite a bit more space and probably further research, when it affects more than two lines. But an initial “ common sense” approach, for junctions with more than two lines, of unifying the intersection points in one node, if they are sufficiently close and within a “convex” space, seems to work fine.







> El 27/10/2014, a las 9:47, SUBSCRIBE SPACESYNTAX Anonymous <[log in to unmask]> escribió:
> 
> Thanks Professors Alen, Bin, Juan for your replies on this post. Sorry if I'm asking for bit of technical information on this post, but I would be appreciate if other people also could come on to the board, and mention the process through which they run an 'angular analysis'. For the purpose of this post, let's assume we decide to go with 'angular analysis;. So which steps you take (compared with those I mentioned below). Specially, what type of 'axial lines' map you use (traditional ones based on visibility drawing by HAND, Auto, or the one proposed based on natural roads'. I believe, this step would be important in kind of standardization and avoid arbitrary process in space syntax research (which many pick on this as an obvious weaknesses of our SS research).  
> 
> Thanks