[GRASS-user] r.futures not working

[Helmut Kudrnovsky]

>regarding OSGeo4W-winGRASS, unfortunately there was a change in the rbatch
package recently which >breaks adding R to %PATH%
>
>*start OSGeo4W in advanced mode, type rbatch in the search window on the
top left
>* open the tree of "Command_Utilities" and "Libs" and choose version 149-4
instead of 149-5
>* finish installation
>* open OSGeo4W shell and type R
>* then R should start in OSGeo4W shell 

switched back to rbatch 149-4 here in my OSGeo4W environment.

installed all needed R packages mentioned in the workshop [1]

**R (≥ 3.5.0) - needed for r.futures.potential with packages:
**
**   MuMIn, lme4, optparse, rgrass7

local R version: 4.03

trying the workshop example with r.futures.potential (where R is needed):

r.futures.potential -d input=sampling at user1
output=D:\wd\rfutures\futures_Asheville_files\potential.csv
columns=devpressure_0_5_92,slope,road_dens_perc,forest_1992_smooth_perc,dist_to_water_km,dist_to_protected_km,dist_interchanges_km,travel_time
developed_column=urban_change_clip subregions_column=counties
min_variables=4

Running automatic model selection ... <=

addon invoked RScript for model selection correctly; waiting for calculation
be finished:

WARNING: Beim Start - Warnmeldungen:
1: Setting LC_CTYPE=de_AT.cp1252 failed
2: Setting LC_COLLATE=de_AT.cp1252 failed
3: Setting LC_TIME=de_AT.cp1252 failed
4: Setting LC_MONETARY=de_AT.cp1252 failed
Fixed term is "(Intercept)"
Warnmeldung:
In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
Model failed to converge with max|grad| = 0.00403567 (tol = 0.002, component
1)
Best model summary:
-------------------------
Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]
Family: binomial  ( logit )
Formula: urban_change_clip ~ devpressure_0_5_92 + dist_interchanges_km +
dist_to_protected_km + dist_to_water_km + forest_1992_smooth_perc +
road_dens_perc + slope + travel_time + (1 | counties)
Data: input_data

AIC      BIC   logLik deviance df.resid
2641.7   2708.6  -1310.8   2621.7     5916

Scaled residuals:
Min      1Q  Median      3Q     Max
-5.5600 -0.2166 -0.0876 -0.0466 21.1719

Random effects:
Groups   Name        Variance Std.Dev.
counties (Intercept) 0.04751  0.218
Number of obs: 5926, groups:  counties, 5

Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept)              0.952166   0.283639   3.357 0.000788 ***
devpressure_0_5_92       0.030397   0.004947   6.144 8.05e-10 ***
dist_interchanges_km    -0.054705   0.013647  -4.009 6.11e-05 ***
dist_to_protected_km    -0.205176   0.044363  -4.625 3.75e-06 ***
dist_to_water_km        -0.157337   0.025512  -6.167 6.95e-10 ***
forest_1992_smooth_perc -0.044996   0.002423 -18.572  < 2e-16 ***
road_dens_perc           0.080876   0.007976  10.140  < 2e-16 ***
slope                    0.024958   0.009225   2.705 0.006822 **
travel_time             -4.057411   0.439160  -9.239  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Correlation of Fixed Effects:
(Intr) d_0_5_ dst_n_ dst_t_p_ dst_t_w_ f_1992 rd_dn_ slope
dvpr_0_5_92 -0.259
dst_ntrchn_ -0.360  0.284
dst_t_prtc_ -0.303  0.099  0.156
dst_t_wtr_k -0.203  0.119 -0.101 -0.100
frst_1992__ -0.256  0.158 -0.077  0.045    0.085
rod_dns_prc -0.378 -0.355 -0.008 -0.094   -0.063    0.057
slope       -0.151  0.058  0.061 -0.006   -0.093   -0.520 -0.009
travel_time -0.558 -0.116  0.029  0.036    0.014    0.101  0.167 -0.013
(Fri Dec 11 23:14:30 2020) Befehl ausgeführt (9 Min 3 Sek)                      

findings:

- there is some locale issue: Setting LC_CTYPE=de_AT.cp1252 failed - it's a
german windows 10 box here
- In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  : <=
not sure about this one here
- Model failed to converge with max|grad| = 0.00403567 (tol = 0.002,
component 1) <= maybe related to input data (not checked if these make
sense)
- Best model summary:
-------------------------
Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]

though it seems a best model could be found

to sum up, r.futures.potential invoking some R scripts works here in OSGeo4W
using rbatch 149-4 

[1]
https://grasswiki.osgeo.org/wiki/Workshop_on_urban_growth_modeling_with_FUTURES#Workshop_data





-----
best regards
Helmut
--
Sent from: http://osgeo-org.1560.x6.nabble.com/Grass-Users-f3884509.html