Re: [UrbanSim-Users] estimating development_project_location_choice_model

No, the DPLCM does not use the land price implicitely as an independent
variable. We just assume that the user will use it (since the price of
land should be a decision factor for developing real estate) and
therefore by default we turn the LPM on prior to the DPLCM estimation.
If user does not include a price variable in the specification, the LPM
can be turned off. But leaving it on, won't change the results, only
delay the estimation.

Hana

Brian Miles wrote:
>
> Thanks Hana. Adding development_constraints to our config and
> recreating the baseyear cache did the trick.
>
> You mentioned that DPLCM implicitly uses land prices as an independent
> variable--i.e. whether or not one has chosen to specify such a
> relationship in their model specification--hence the need to run the
> land price model as part of estimating the DPLCM. (you said that this
> relationship can be turned off in
> urbansim/configs/dplcm_estimation_config.py, method
> get_dplcm_configuration.)
>
> I don't see any mention of this relationship in section 8.4.5 of the
> urbansim manual (pp. 111ff) where the
> development_project_location_choice_model is discussed. Are there any
> other models which rely on such implicit relationships? If so, if these
> documented in the manual, I would like to request that they be.
>
> Thanks for all of your help,
>
> Brian
>
> On Jan 11, 2008, at 5:32 PM, Hana Sevcikova wrote:
>
>> Here is the cause (and solution) to this problem:
>>
>> When estimating the DPLCM, the code needs to determine capacity in
>> previous years. Thus, it needs a few datasets available in the cache
>> of the previous years. Those datasets are defined in the entry
>> tables_to_copy_to_previous_years of your configuration. It turned out
>> that the eugene configuration is missing one table here
>> (development_constraints) which causes your error. I have updated it
>> in the repository. The correct entry looks like this:
>>
>> tables_to_copy_to_previous_years = {
>> 'development_type_group_definitions': 1975,
>> 'development_type_groups': 1975,
>> 'development_types': 1975,
>> 'development_constraints':1975,
>> 'urbansim_constants': 1975,
>> },
>>
>> You can either modify your configuration file and rerun the script
>> opus_core/tools/create_baseyear_cache.py. If the entry cache_directory
>> points to an existing cache nothing will be cached, but gridcells will
>> be unrolled again and the correct datasets copied to the previous years.
>>
>> Alternatively, just copy the directory development_constraints from
>> your base year cache to the earliest year you have data for.
>>
>> I hope this will work.
>>
>> Hana
>>
>>
>> Brian Miles wrote:
>>> Hi Hana,
>>> We checked this out and in the process built a simplified
>>> development_constraints table (see below) that we hoped would work
>>> for now, but sadly it did not.
>>> What else should we look at?
>>> Also, we don't understand well why before the estimation of
>>> dplcm-residential, for which we've only specified one variable
>>> (urbansim.gridcell.ln_residential_units) for our current testing, the
>>> land price model is being run. I suppose it's run before the
>>> estimation because land prices are a criterion for dplcm, but I'm
>>> wondering where this relationship is defined if not in the dplcm
>>> model specification.
>>> Thanks,
>>> Brian
>>> mysql> select * from development_constraints;
>>> +---------------+---------+---------------+-----------+-----------+-------------
>>> --------+---------------------+---------------------+---------------------+
>>>
>>> | CONSTRAINT_ID | CITY_ID | IS_IN_WETLAND | MIN_UNITS | MAX_UNITS |
>>> MIN_COMMERCIAL_SQFT | MAX_COMMERCIAL_SQFT | MIN_INDUSTRIAL_SQFT |
>>> MAX_INDUSTRIAL_SQFT |
>>> +---------------+---------+---------------+-----------+-----------+---------------------+---------------------+---------------------+---------------------+
>>> | 1 | -1 | 1 | 0 | 10
>>> | 0 | 1000 | 0 |
>>> 10000 |
>>> +---------------+---------+---------------+-----------+-----------+---------------------+---------------------+---------------------+---------------------+
>>> 1 row in set (0.00 sec)
>>> On Jan 8, 2008, at 1:19 PM, Hana Sevcikova wrote:
>>>> Brian,
>>>>
>>>> My suspicion is that there is something wrong with the column names
>>>> of your development_constraints table. Could you check please that
>>>> the table header corresponds to that described in
>>>> http://www.urbansim.org/docs/opus-userguide/node251.html
>>>>
>>>> Thanks,
>>>>
>>>> Hana
>>>>
>>>>
>>>> Brian Miles wrote:
>>>>> Hello,
>>>>> I'm trying to estimate the
>>>>> residential_development_project_location_choice_model. I'm
>>>>> getting the following error in the compute method of
>>>>> total_maximum_development_SSS.py. Before I spent too much time
>>>>> digging, I wanted to see if anyone had encountered something like
>>>>> this.
>>>>> Thanks,
>>>>> Brian Miles
>>>>> Graduate Assistant
>>>>> Spatial Analysis Lab - Rubenstein School of Environment and
>>>>> Natural Resources
>>>>> University of Vermont
>>>>> excerpt from stack trace:
>>>>> File "C:\opusworkspace-svn\urbansim\gridcell
>>>>> \total_maximum_development_SSS.py"
>>>>> , line 40, in compute
>>>>> index=where(is_developable)[0])[self.type][:,
>>>>> 1]).astype(self._return_type)
>>>>> File
>>>>> "C:\opusworkspace-svn\urbansim\datasets\gridcell_dataset.py", line
>>>>> 56, in
>>>>> get_development_constrained_capacity
>>>>> constraints.get_attribute_names(), ["min", "max"])
>>>>> File "C:\opusworkspace-svn\opus_core\misc.py", line 325, in
>>>>> remove_elements_wi
>>>>> th_matched_prefix_from_list
>>>>> m = apply_along_axis(match_prefix, 1, pl.reshape(pl.size,1),
>>>>> array(valuelist
>>>>> ))
>>>>> File "C:\Python25\lib\site-packages\numpy-1.0.3-py2.5-win32.egg
>>>>> \numpy\lib\shap
>>>>> e_base.py", line 30, in apply_along_axis
>>>>> res = func1d(arr[tuple(i.tolist())],*args)
>>>>> File "C:\opusworkspace-svn\opus_core\misc.py", line 320, in
>>>>> match_prefix
>>>>> t = apply_along_axis(match_single_element, 1,
>>>>> arr.reshape(arr.size,1), prefi
>>>>> x[0])
>>>>> File "C:\Python25\lib\site-packages\numpy-1.0.3-py2.5-win32.egg
>>>>> \numpy\lib\shap
>>>>> e_base.py", line 30, in apply_along_axis
>>>>> res = func1d(arr[tuple(i.tolist())],*args)
>>>>> IndexError: invalid index
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> Users_at_urbansim.org
>>>>> http://www.urbansim.org/mailman/listinfo/users
>
Received on Fri Jan 11 2008 - 16:11:45 PST