Re: [UrbanSim-Users] travel_data

Hi,
 
I don't quite understand why no alteration is necessary to use travel
time data instead of logsums. It seems that we would at least need to
multiply the travel times by -1. For example, in the file
access_to_workplace_from_residences_DDD.py, access is defined as
SUM(Households(i) * exp(logsum_DDD(i to j)), for i=zone_1...zone_n), and
just putting travel times into the travel data file won't give us the
correct results.

Jen

Paul Waddell wrote:
> Essentially, you just need to put into the table travel_data, a
> from_zone, to_zone, and whatever travel model skim variables you want to
> use. We use travel times in the am peak by drive alone, transit walk,
> and walk modes, in some applications, to compute the number of jobs
> reachable within 30 minutes, for example. We also use generalized
> costs, if available. Numerous variables are already defined that use
> these times and generalized costs, in the psrc and semcog applications,
> for example.
>
> I'll look into putting these applications into the installer we are
> going to release shortly.
>
> Paul
>
> Jen Duthie wrote:
>
>> Hi,
>>
>> Can someone please help me figure out how to alter the code to use
>> travel times instead of logsums? Exactly which files need to be
>> altered, and how should they be altered?
>>
>> Thanks,
>> Jen
>>
>>
>> Paul Waddell wrote:
>>
>>
>>> UrbanSim is generally connected to an existing regional travel model,
>>> from which either composite utilities from the mode choice model (aka
>>> logsums) or more generally, congested travel times by mode for the
>>> home-based-work purpose in the a.m. peak period (or other similar
>>> measures) are extracted for each travel model year. The connections can
>>> be automated - we have done this with emme/2 and transcad most recently,
>>> and previously with TP+ and Minutp. It essentially requires an
>>> interface to be written for whichever travel model you are using, to
>>> summarize UrbanSim output and generate the inputs required for the
>>> travel model; and in the other direction, to extract travel model
>>> outputs and generate the travel_data table -- with whatever travel
>>> measures you want to use. Logsums are by no means required, and we have
>>> found them to be somewhat unstable in practice (though they look good in
>>> theory) - with small changes in the travel model causing large shifts in
>>> the logsum measures and therefore precipitating a recalibration of UrbanSim.
>>>
>>>
>>>
>>>
>> _______________________________________________
>> Users mailing list
>> Users_at_urbansim.org
>> http://www.urbansim.org/mailman/listinfo/users
>>
>>
> _______________________________________________
> Users mailing list
> Users_at_urbansim.org
> http://www.urbansim.org/mailman/listinfo/users
>
Received on Fri Aug 31 2007 - 11:44:54 PDT