[GRASS-user] r.futures.demand

[Anna Petrášová]

Hi Mitchell,

looking at your command there are several things to fix. Parameter columns
should have only the predictors, so for example remove county_rast and
urban_change_01_16. Parameter developed_column should be the response
variable, so I suppose urban_change_01_16 is the one to use here. Lastly,
without the dredge mode only use combination of predictors that are not
correlated and don't represent the same info, for example, do not include
dist_toh20_2001_km,dist_toh20_2004_km,dist_toh20_2006,dist_toh20_2008_km,dist_toh20_2011_km,dist_toh20_2013_km,dist_toh20_2016_km,
since these all represent the same variable (I suppose). Without the dredge
mode, all provided predictors will be used in the model, with the dredge
mode, it will try different combinations and return the best model.

Note r.futures.potential is basically just a wrapper around R glmer (lme4)
function, so if you are familiar with R, you can experiment with finding a
model directly in R if you need more flexibility etc.

Please keep the conversation on this mailing list which is archived, so
that others can add to it or benefit from it later.

Best,
Anna

On Thu, Feb 11, 2021 at 2:12 PM Mitchell Meads <mitchell.meads at tamu.edu>
wrote:

> Hi Anna,
>
> I've revisited the r.futures.potential submodel for the futures modeling
> and I still can't get the r.futures.potential submodel to run. I've
> rescaled my variables so they all should be on a similar scale but the
> program is telling me that the variables are still not in appropriate scale
> along with some other error messages that I have copied below. Please let
> me know what you think I'm doing wrong.
>
> (Thu Feb 11 13:11:21 2021)
>
> r.futures.potential --overwrite input=sampling_f_km at PERMANENT
> output=potential_f_km.csv
> columns=devclip_2016,block_rast,county_rast,devpress_0_5_2016,dist_to_evac_km,dist_to_pp_km,dist_toh20_2001_km,dist_toh20_2004_km,dist_toh20_2006,dist_toh20_2008_km,dist_toh20_2011_km,dist_toh20_2013_km,dist_toh20_2016_km,per_forest_2001,per_forest_2004,per_forest_2006,per_forest_2008,per_forest_2011,per_forest_2013,per_forest_2016,urban_change_01_16
> developed_column=devclip_2016 subregions_column=county_rast min_variables=5
> max_variables=10
> Computing model...
> WARNING: fixed-effect model matrix is rank deficient so dropping 2 columns
> / coefficients
> Error in pwrssUpdate(pp, resp, tol = tolPwrss, GQmat = GQmat, compDev =
> compDev,  :
> pwrssUpdate did not converge in (maxit) iterations
> Calls: glmer ... <Anonymous> -> <Anonymous> -> stopifnot -> fn ->
> pwrssUpdate
> In addition: Warning message:
> Some predictor variables are on very different scales: consider rescaling
> Execution halted
> WARNING: fixed-effect model matrix is rank deficient so dropping 2 columns
> / coefficients
> Error in pwrssUpdate(pp, resp, tol = tolPwrss, GQmat = GQmat, compDev =
> compDev,  :
> pwrssUpdate did not converge in (maxit) iterations
> Calls: glmer ... <Anonymous> -> <Anonymous> -> stopifnot -> fn ->
> pwrssUpdate
> In addition: Warning message:
> Some predictor variables are on very different scales: consider rescaling
> Execution halted
> ERROR: Running R script failed, check messages above
> (Thu Feb 11 13:11:25 2021) Command finished (3 sec)
>
> Cheers,
> Mitchell Meads, PhD Student
> Texas A&M University at Galveston
> Marine & Coastal Management Sciences; CTBS <http://www.tamug.edu/ctbs/>
> Email: mitchell.meads at tamu.edu
>
>
>
>
> On Sat, Jan 23, 2021 at 9:54 PM Anna Petrášová <kratochanna at gmail.com>
> wrote:
>
>> Hi Mitchell,
>>
>> On Fri, Jan 22, 2021 at 3:23 PM Mitchell Meads <mitchell.meads at tamu.edu>
>> wrote:
>>
>>> Hi all,
>>>
>>> I'm getting an error when running r.futures.demand that says "Number of
>>> development raster maps doesn't not correspond to the number of observed
>>> times". I thought that a fix for this would be to make sure my input
>>> population csv had years that matched the input raster maps in the first
>>> column but this didn't change the error. If you could let me know what you
>>> think I'm doing wrong, that would be greatly appreciated!
>>>
>>
>> The error means the number of raster maps in the development parameter
>> does not match the number of years in the observed_population CSV. So for
>> example, if you specify development=urban_2001,urban_2006,urban_2011 (3
>> maps) then the observed_population file should look like this (header + 3
>> lines):
>>
>> year,37037,37063,...
>> 2001,19860,10980,...
>> 2006,20760,12660,...
>> 2011,21070,13090,...
>>
>> If you can't see any obvious mistake with your data, I would have to see
>> the r.futures.demand command you are trying to execute and the input CSV
>> file.
>>
>>
>>>
>>>
>>> Cheers,
>>> Mitchell
>>>
>>>
>>> _______________________________________________
>>> grass-user mailing list
>>> grass-user at lists.osgeo.org
>>> https://lists.osgeo.org/mailman/listinfo/grass-user
>>> <https://urldefense.com/v3/__https://lists.osgeo.org/mailman/listinfo/grass-user__;!!KwNVnqRv!S-XTr0kWrbJ704p6CIighh6PUpDBJchKEjSQMopVh0bXjHxBYSqNzjaTNlgopVQ1ulKp$>
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-user/attachments/20210211/1163506c/attachment.html>