Mona,

the simplest advice is that if you can think of two ways of doing an analysis you do both and compare the results. For example, the void cannot be walked across so treat it as a block for permeability, but it can be seen across so treat it as open space for visibility. The difference between these two analyses will tell you something about how the building works.

Alan

> On 19 Oct 2018, at 12:59, Mona Tarashi <[log in to unmask]> wrote:
> 
> Dear friends
> Hello
> 
> I have a problem in performing an analysis. we have void in a plan.
> so we can see the first floor from the second floor, but we have no direct access. ( just from staircases)
> i want to perform axial analysis and VGA analysis, but i am not sure it is correct not to consider void areas.
> do you have any comments? or any similar project which i can see the process?
> 
> Thanks in advance
> 
> 
> On Fri, Oct 19, 2018 at 3:17 PM Jacob Dibble <[log in to unmask]> wrote:
> Dear Kimon, Daniel, and Alan,
> 
> Thank you all for the quick and thorough responses!! I am actually going to hold off on any further responses to you because I want to look more carefully at your answers, run a few different test, and revisit the formulas used of course in conjunction with what you have said. Also, Kimon I want to check out your publication and Daniel I would like to check out the reference you sent me as well.
> 
> I will come back when I have a chance to look more calmly at your very thorough responses and go from there!
> 
> Thanks again and hope everyone has a nice weekend,
> Jacob
> 
> --
> Dr. Jacob Leonard Dibble
> 
> www.jacobdibble.com
> 
> [log in to unmask]
> +358 (0) 45 7877 9956
> 
> 
> On Fri, Oct 19, 2018 at 12:42 PM Krenz, Kimon-Vincent <[log in to unmask]> wrote:
> Dear Jacob,
> 
> I assume your are running an angular segment analysis with metric radius for integration and are using a road centre line map as your base?.
> 
> I have discussed this issue a while ago with Petros Koutsolampros. The conclusion is that these outliers are caused by three factors: 
> 
> a) the length of the segment, 
> b) the fact that segments are leaf segments (at the end or near the end of a graph), and 
> c) the use of a tulip analysis (which divides the angular difference into tulip bins).
> 
> This means the analysis only reaches the segment only once and this leads to high square values. Petros wrote a simple example based on an analysis back then, which I will copy below:
> 
> Taking the example of the outlier Ref: 6001. That segment is only connected to one other segment (Ref: 5999) and because you limit by metric radius 1200, you only capture that for the analysis from 6001. You can see this if you do metric-step-depth: only 5999 is less than 1200.
> 
> The actual angle between the two segments is 178.44057 degrees but since you're using tulip analysis with 1024 bins this falls within the 4th bin (1st bin is at 180 if I understand correctly), making thus their angular distance = bin/(tulips / 4) = 1/64 (0.015625). Again you can see this with angular step depth from 6001: 5999 has that value.
> 
> As there's only one other segment, this is also the final total depth from 6001
> 
> The final integration calculation is nodes*nodes/totalDepth = 2*2/(1/64) = 256 (the two nodes are 6001 and 5999)
> 
> As one can see, this is an expected outcome, at least from a computational point of view. 
> 
> You might, nevertheless, want to remove these outliers from your map. If so, you could have a look at the appendix (page 74.23) of my SSS12 paper 'regional morphology', where I describe a method to identify these outlier by dividing through the log of integration by the log total depth: 
> 
> CCTD_r = Log(CC_r+3)/Log(TD_r+3)
> 
> https://www.researchgate.net/publication/322255338_Regional_Morphology_The_Emergence_of_Spatial_Scales_in_Urban_Regions
> 
> Best,
> Kimon
> 
> 
> Dr Kimon Krenz
> FHEA
> 
> mail.       [log in to unmask]
> phone.    0044 7784 329089
> web.       www.kimonkrenz.com
> 
> The Bartlett School of Architecture
> Faculty of the Built Environment
> University College London UCL
> 22 Gordon Street
> London WC1H 0QB
> 
> 
> 
>> On 19 Oct 2018, at 09:34, Jacob Dibble <[log in to unmask]> wrote:
>> 
>> Dear All,
>> 
>> I had sent this query to the QGIS toolkit forum and Jorge advised me to instead ask on the general forum:
>> 
>> I have a question regarding Local Integration values on smaller radii. On low radii, on even higher (up to 800m) I am seeing in my particular case that the segments with the highest integration values are on the very very ends of the graph. The attached picture shows the 400m integration for segment analysis with metric radius type and only the top decile of highest values shown in red, with the base network underneath. For example one of the segments at the very edge of the graph has an integration value about 8 times higher than the next highest.
>> 
>> I am looking for an explanation and some help to understand why some of the highest values are appearing at the ends of the graph, or in the middle of the arterial routes coming towards the city and even in the centre. These segments really only reach one or two segments within the 400m cutoff, so logically these should have very low integration, right?
>> 
>> It would be nice to discuss perhaps why this is happening (both technically and perhaps from a perspective related to the urban structure) and also to know if there are any suggestions, workarounds, ways to redraw the graph, etc. 
>> 
>> Thanks in advance,
>> Jacob Dibble
>> 
>> <image.png>
>> 
>> To unsubscribe from the SPACESYNTAX list, click the following link:
>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SPACESYNTAX&A=1
>> 
> 
> 
> To unsubscribe from the SPACESYNTAX list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SPACESYNTAX&A=1
> 
> 
> To unsubscribe from the SPACESYNTAX list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SPACESYNTAX&A=1
> 
> 
> 
> -- 
> Mona Tarashi.
> Master of urban design
> Tehran Art University
> 
> To unsubscribe from the SPACESYNTAX list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SPACESYNTAX&A=1
> 

########################################################################

To unsubscribe from the SPACESYNTAX list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SPACESYNTAX&A=1