Hans, See my comments below.
> HSP>>> Sure the sw is inessential for the present application. I was
> only curious know if you had managed to program ABM capabilities into
> ArcGIS (which as far as I know only has the shallow
> 'AgtenAnalyst/Repast' package this far) or how far you have managed to
> take NetLogo in terms of networks (which in its GIS-extension dosen't
> facilitate network representation/analysis.
In fact we have implemented car-car interaction and car-street
interaction in both ArcGIS and NetLogo. They are pretty straightforward.
>
> > 2) Is the NetLogo model-code you used for agents interaction with the
> > network publicly available?
> Yes.
> HSP>>> How can I get my hands on it?
We consider to put it publicly once the paper is published. However,
algorithms are clearly described in the paper.
> HSP>>> I appears that you have three 'sets' of information: Real-world
> counts (Taxies in London), network metrics, and agent-simulations. A
> main motivation - as per you introduction - is to investigate which
> networks metric (classic SS + two Rank-based) explains the actual - or
> simulated - traffic flows best. The conclusion is rightfully that the
> flow to a high extent can be explained by means of metrics. Since
> simulated and actual flows are correlated, this means that metrics can
> be used as proxies for (relative) flow patterns (parameter estimates
> can be used to 'upscale' metric indicators to real-world flows).
> Accordingly, this can (if you are putting your glasses in the right
> (or wrong) angle) be read as there's no need for all the fuzz of
> setting up an ABM if flow patterns is the ultimate goal any way.
yes, now I am with you. Simply traffic flow can be predicted by the
metrics.
> HSP>>> I was just wondering if you could elaborate on why ABM are
> needed (which I of course see that they are) in spite of our access to
> such powerful indicators - based on metrics - of flow..
ok, see your point, to setup a "would-be" world which you can experiment on.
> HSP>>> As I see it, the physical environment (in particular the
> network) is the scene on which humans (and other agents) act. The
> environment doesn't predict or dictate. By means of better descriptors
> of the environment - for instance network metrics - we can describe
> the 'scene' (e.g. in terms of 'background' or 'average flows') of our
> agents and thereby make more realistic simulations of the few,
> detailed behaviours in focus. The study of such indicators - like
> yours - is of course of great interest in this respect.
Certainly the simulations can be refined as you suggested.
Thanks and cheers.
Bin
--
--------------------------------------------------------
Bin Jiang
Division of Geomatics, KTH Research School
Department of Technology and Built Environment
University of Gävle, SE-801 76 Gävle, Sweden
Phone: +46-26-64 8901 Fax: +46-26-64 8828
Email: [log in to unmask] Web: http://fromto.hig.se/~bjg/
--------------------------------------------------------
European Associate Editor
Computers, Environment and Urban Systems: An International Journal
NordGISci: http://fromto.hig.se/~bjg/NordGISci/
ICA Commission: http://fromto.hig.se/~bjg/ica/
|