<div dir="ltr"><div>Hi Moritz,<br></div><div><br></div>Thanx for the advice! Will check it out. From the publications I read, I inferred it was better to use an alternative method (data mining/ object-based) on high resolution imagery than pixel-based. I also read that texture metrics (e.g. standard deviation of spectral bands) may
<span style="font-size:small;text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">capture the unique “patterns”</span> of wetlands and other ecosystems (given the appropriate spatial scale). Will continue testing and checking for improvements.<div><div><br></div><div>I was running directy from GRASS GIS, but recently created and used a new training map with the lower case column name. My apologies for the inconsistency, but good to know that the previous training map wasn't the issue. </div></div><div><br></div><div>Best,</div><div>Jamille</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Jun 19, 2018 at 6:58 AM, Moritz Lennert <span dir="ltr"><<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldonline.be</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">P.S. I'm not sure that for the landscape you are working on (mainly forest areas) OBIA is the best approach. You might want to also try a pixel-based approach with i.smap:<br>
<br>
#create raster training areas<br>
v.to.rast Training_Ben5 out=training use=attr attrcol=Ecosystem<br>
#create a subgroup with all four bands<br>
i.group WV_Benatimofo subg=sub in=WV_Benatimofo.1@haarlooj_Be<wbr>n_Test,WV_Benatimofo.2@haarloo<wbr>j_Ben_Test,WV_Benatimofo.3@<wbr>haarlooj_Ben_Test,WV_<wbr>Benatimofo.4@haarlooj_Ben_Test<br>
#create signature file<br>
i.gensigset training group=WV_Benatimofo sub=sub<br>
#classify<br>
i.smap group=WV_Benatimofo sub=sub sig=sig out=smap_classified goodness=smap_goodness<span class="HOEnZb"><font color="#888888"><br>
<br>
Moritz</font></span><div><div class="h5"><br>
<br>
On 19/06/18 11:33, Moritz Lennert wrote:<br>
</div></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div class="h5">
Hi Jamille,<br>
<br>
Testing with the data you sent me offlist (which BTW was a mapset, not a<br>
location. A location contains at least one mapset named PERMANENT which<br>
contains the projection info - I just assumed that you are working in<br>
UTM Zone 21N, created my own location and copied your mapset into that<br>
location.) and with the following command:<br>
<br>
v.class.mlR -i --overwrite<br>
segments_map=Segments_vector_S<wbr>tats_Ben_test@haarlooj_Ben_Tes<wbr>t<br>
training_map=Training_Ben5@haa<wbr>rlooj_Ben_Test<br>
raster_segments_map=best5_myre<wbr>gion1_at_haarlooj_Ben_Test_<wbr>rank1@haarlooj_Ben_Test<br>
train_class_column=Ecosystem output_class_column=vote<br>
output_prob_column=prob classifiers=svmRadial,rf,C5.0 folds=5<br>
partitions=10 tunelength=10 weighting_metric=accuracy<br>
r_script_file=R-script processes=3<br>
<br>
the module runs perfectly fine for me here on GNU/Linux. Note the fact<br>
that the train_class_column=Ecosystem, not ecosystem. Case matters here.<br>
<br>
Could you try with the training class column in the right case ?<br>
<br>
Moritz<br>
<br>
On 18/06/18 21:51, Jamille Haarloo wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hello Moritz,<br>
<br>
It looks like I got some results, but I suspect there are still some<br>
issues due to the warning messages.<br>
<br>
I either kept getting that a file couldn't be found or that it had<br>
trouble running the R-script.<br>
My actions:<br>
<br>
1. I found that it was looking in "C:\Program Files\QGIS<br>
2.16.0\apps\Python27\Lib" for certain scripts but I also had another<br>
location of the Python27 library "C:\Program Files\GRASS GIS<br>
7.4.0\Python27\Lib". So I tried adding the second location via the<br>
black terminal because I figured it needed the GRASS<br>
versions/formats (with my lack of experience I am not sure if I<br>
succeeded).<br>
2. It was still failing, and I suspected this had to do with the qbwwv<br>
voting method (see output). So I unchecked that option and got<br>
results. I'am also sending a screenshot of the attribute table with<br>
results.<br>
<br>
Do you have any suggestions for improvement?<br>
Soon I will test with a much bigger area for a land-use planning project.<br>
<br>
<br>
/The last 2 command outputs:/<br>
<br>
Running R now. Following output is R output.<br>
During startup - Warning messages:<br>
1: Setting LC_CTYPE=en_US.cp1252 failed<br>
2: Setting LC_COLLATE=en_US.cp1252 failed<br>
3: Setting LC_TIME=en_US.cp1252 failed<br>
4: Setting LC_MONETARY=en_US.cp1252 failed<br>
Loading required package: caret<br>
Loading required package: lattice<br>
Loading required package: ggplot2<br>
Loading required package: foreach<br>
Loading required package: iterators<br>
Loading required package: parallel<br>
During startup - Warning messages:<br>
1: Setting LC_CTYPE=en_US.cp1252 failed<br>
2: Setting LC_COLLATE=en_US.cp1252 failed<br>
3: Setting LC_TIME=en_US.cp1252 failed<br>
4: Setting LC_MONETARY=en_US.cp1252 failed<br>
During startup - Warning messages:<br>
1: Setting LC_CTYPE=en_US.cp1252 failed<br>
2: Setting LC_COLLATE=en_US.cp1252 failed<br>
3: Setting LC_TIME=en_US.cp1252 failed<br>
4: Setting LC_MONETARY=en_US.cp1252 failed<br>
Warning message:<br>
In nominalTrainWorkflow(x = x, y = y, wts = weights, info = trainInfo, :<br>
There were missing values in resampled performance measures.<br>
Error in `$<-.data.frame`(`*tmp*`, vote_qbwwv, value = numeric(0)) :<br>
replacement has 0 rows, data has 1965<br>
Calls: $<- -> $<-.data.frame<br>
Execution halted<br>
ERROR: There was an error in the execution of the R script.<br>
Please check the R output.<br>
(Mon Jun 18 15:36:47 2018) Command finished (1 min 24 sec)<br>
(Mon Jun 18 15:50:30 2018)<br>
v.class.mlR -i --overwrite<br>
segments_map=Segments_vector_S<wbr>tats_Ben_test@haarlooj_Ben_Tes<wbr>t<br>
training_map=Training_Ben_Gras<wbr>s@haarlooj_Ben_Test<br>
raster_segments_map=best5_myre<wbr>gion1_at_haarlooj_Ben_Test_<wbr>rank1@haarlooj_Ben_Test<br>
train_class_column=ecosystem output_class_column=vote<br>
output_prob_column=prob classifiers=svmRadial,rf,C5.0 folds=5<br>
partitions=10 tunelength=10 weighting_metric=accuracy<br>
r_script_file=C:\Users\haarloo<wbr>j\Documents\CELOS\v.class.<wbr>mIRR_optional_output\Ben_test_<wbr>mlR-script3<br>
processes=2<br>
Running R now. Following output is R output.<br>
During startup - Warning messages:<br>
1: Setting LC_CTYPE=en_US.cp1252 failed<br>
2: Setting LC_COLLATE=en_US.cp1252 failed<br>
3: Setting LC_TIME=en_US.cp1252 failed<br>
4: Setting LC_MONETARY=en_US.cp1252 failed<br>
Loading required package: caret<br>
Loading required package: lattice<br>
Loading required package: ggplot2<br>
Loading required package: foreach<br>
Loading required package: iterators<br>
Loading required package: parallel<br>
During startup - Warning messages:<br>
1: Setting LC_CTYPE=en_US.cp1252 failed<br>
2: Setting LC_COLLATE=en_US.cp1252 failed<br>
3: Setting LC_TIME=en_US.cp1252 failed<br>
4: Setting LC_MONETARY=en_US.cp1252 failed<br>
During startup - Warning messages:<br>
1: Setting LC_CTYPE=en_US.cp1252 failed<br>
2: Setting LC_COLLATE=en_US.cp1252 failed<br>
3: Setting LC_TIME=en_US.cp1252 failed<br>
4: Setting LC_MONETARY=en_US.cp1252 failed<br>
Warning message:<br>
In nominalTrainWorkflow(x = x, y = y, wts = weights, info = trainInfo, :<br>
There were missing values in resampled performance measures.<br>
Finished running R.<br>
Loading results into attribute table<br>
(Mon Jun 18 15:51:58 2018) Command finished (1 min 27 sec)<br>
<br>
<br>
Best,<br>
Jamille Haarloo<br>
Department of Natural Resources and Environmental Assessment (NARENA)<br>
Centre for Agricultural Research in Suriname (CELOS)<br>
<<a href="http://www.celos.sr.org/" rel="noreferrer" target="_blank">http://www.celos.sr.org/</a>><br>
<br>
Prof. Dr. Ir. Jan Ruinardlaan<br>
AdeKUS University campus<br>
Phone: 439982<br>
<br>
<br>
On Thu, Jun 14, 2018 at 11:02 AM, Jamille Haarloo <<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><br>
<mailto:<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a>><wbr>> wrote:<br>
<br>
Hello Moritz,<br>
<br>
Sorry about that. I am sending the zip of the location. The output<br>
at the other machine (system cannot find the file specified) was due<br>
to location/ directory issues and has been resolved, but also stops<br>
executing the R-script saying there were more than 50 errors. Been<br>
looking for similar cases online but haven't found many.<br>
<br>
<br>
haarlooj_Ben_Test.zip<br>
<<a href="https://drive.google.com/file/d/1wQqmreMbbd3oRfoVXWIIGJOG9L63miV6/view?usp=drive_web" rel="noreferrer" target="_blank">https://drive.google.com/fil<wbr>e/d/1wQqmreMbbd3oRfoVXWIIGJOG9<wbr>L63miV6/view?usp=drive_web</a>><br>
<br>
<br>
On Thu, Jun 14, 2018 at 5:19 AM, Moritz Lennert<br>
<<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldonline.be</a> <mailto:<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldonl<wbr>ine.be</a>>><br>
wrote:<br>
<br>
P.S. Your vector data also does not contain any attribute<br>
information as this is contained in a separate sqlite file. Just<br>
another reason never to touch the contents of a GRASS database<br>
directly, but only using dedicated tools. :-)<br>
<br>
<br>
On 14/06/18 10:11, Moritz Lennert wrote:<br>
<br>
Hi Jamille,<br>
<br>
Sorry, but I've been hopping from meeting to meeting these<br>
days and<br>
haven't had the opportunity to look at your data.<br>
<br>
Actually, the form you transmit them in is very<br>
inconvenient. If you<br>
want to transmit GRASS data to someone, you should either<br>
zip an entire<br>
location directory and send that, or export the data with<br>
r.pack/v.pack.<br>
<br>
In this form it is difficult to know which projection is<br>
used and I have<br>
to handcraft the data by copying the files into different<br>
directories.<br>
<br>
So, please resend your data in a format that I can easily use.<br>
<br>
Best wishes,<br>
Moritz<br>
<br>
On 11/06/18 20:55, Jamille Haarloo wrote:<br>
<br>
Hello Moritz,<br>
<br>
I managed to edit my copy of v.class.mlR with the lines<br>
you sent. But I<br>
am still getting errors on two different machines.<br>
I am sending you the data (hope I didn't miss anything)<br>
and the command<br>
outputs. I think there is sth wrong with my computer,<br>
but also with one<br>
of the files because of the error on the other machine.<br>
<br>
_Computer I normally use:_<br>
<br>
v.class.mlR -i --overwrite<br>
segments_map=Segments_vector_<wbr>Stats_Ben_test@haarlooj_Ben_Te<wbr>st<br>
training_map=Training_Ben5@ha<wbr>arlooj_Ben_Test<br>
raster_segments_map=best5_myr<wbr>egion1_at_haarlooj_Ben_Test_<wbr>rank1@haarlooj_Ben_Test<br>
train_class_column=Ecosystem output_class_column=vote<br>
output_prob_column=prob classifiers=svmRadial,rf,C5.0<br>
folds=5<br>
partitions=10 tunelength=10 weighting_modes=smv,qbwwv<br>
weighting_metric=accuracy<br>
Running R now. Following output is R output.<br>
During startup - Warning messages:<br>
1: Setting LC_CTYPE=en_US.cp1252 failed<br>
2: Setting LC_COLLATE=en_US.cp1252 failed<br>
3: Setting LC_TIME=en_US.cp1252 failed<br>
4: Setting LC_MONETARY=en_US.cp1252 failed<br>
Loading required package: caret<br>
Loading required package: lattice<br>
Loading required package: ggplot2<br>
There were 50 or more warnings (use warnings() to see<br>
the first 50)<br>
There were 50 or more warnings (use warnings() to see<br>
the first 50)<br>
Error in `$<-.data.frame`(`*tmp*`, vote_qbwwv, value =<br>
numeric(0)) :<br>
replacement has 0 rows, data has 1965<br>
Calls: $<- -> $<-.data.frame<br>
Execution halted<br>
ERROR: There was an error in the execution of the R script.<br>
Please check the R output.<br>
<br>
<br>
_The output from the other machine:_<br>
(Mon Jun 11 13:27:39 2018)<br>
v.class.mlR -i --overwrite<br>
segments_map=Segments_vector_<wbr>Stats_Ben_test@haarlooj_Ben_Te<wbr>st<br>
training_map=Training_Ben5@ha<wbr>arlooj_Ben_Test<br>
raster_segments_map=best5_myr<wbr>egion1_at_haarlooj_Ben_Test_<wbr>rank1@haarlooj_Ben_Test<br>
train_class_column=Ecosystem output_class_column=vote<br>
output_prob_column=prob classifiers=svmRadial,rf,C5.0<br>
folds=5<br>
partitions=10 tunelength=10 weighting_metric=accuracy<br>
Running R now. Following output is R output.<br>
Traceback (most recent call last):<br>
File<br>
"C:\Users\HaarlooJ\AppData\Ro<wbr>aming\GRASS7\addons/scri<br>
pts/<a href="http://v.class.mlR.py" rel="noreferrer" target="_blank">v.class.mlR.py</a> <<a href="http://v.class.mlR.py" rel="noreferrer" target="_blank">http://v.class.mlR.py</a>><br>
<<a href="http://v.class.mlR.py" rel="noreferrer" target="_blank">http://v.class.mlR.py</a>>", line 639, in <module><br>
File<br>
"C:\Users\HaarlooJ\AppData\Ro<wbr>aming\GRASS7\addons/scri<br>
pts/<a href="http://v.class.mlR.py" rel="noreferrer" target="_blank">v.class.mlR.py</a> <<a href="http://v.class.mlR.py" rel="noreferrer" target="_blank">http://v.class.mlR.py</a>><br>
<<a href="http://v.class.mlR.py" rel="noreferrer" target="_blank">http://v.class.mlR.py</a>>", line 576, in main<br>
shutil.copy(r_commands, r_script_file)<br>
File "C:\Program Files\GRASS GIS<br>
7.4.0\Python27\lib\<wbr>subprocess.py", line 537, in check_call<br>
retcode = call(*popenargs, **kwargs)<br>
File "C:\Program Files\GRASS GIS<br>
7.4.0\Python27\lib\<wbr>subprocess.py", line 524, in call<br>
return Popen(*popenargs, **kwargs).wait()<br>
File "C:\Program Files\GRASS GIS<br>
7.4.0\Python27\lib\<wbr>subprocess.py", line 711, in __init__<br>
errread, errwrite)<br>
File "C:\Program Files\GRASS GIS<br>
7.4.0\Python27\lib\<wbr>subprocess.py", line 948, in<br>
_execute_child<br>
startupinfo)<br>
WindowsError: [Error 2] The system cannot find the file<br>
specified<br>
(Mon Jun 11 13:27:42 2018) Command finished (3 sec)<br>
<br>
<br>
On Mon, Jun 11, 2018 at 5:47 AM, Moritz Lennert<br>
<<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldonline.be</a><br>
<mailto:<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldon<wbr>line.be</a>><br>
<mailto:<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldon<wbr>line.be</a><br>
<mailto:<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldon<wbr>line.be</a>>>> wrote:<br>
<br>
Hi Jamille,<br>
<br>
Le Fri, 8 Jun 2018 16:14:45 -0300,<br>
Jamille Haarloo <<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><br>
<mailto:<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><wbr>><br>
<mailto:<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><br>
<mailto:<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><wbr>>>> a écrit :<br>
<br>
> Hello Moritz,<br>
><br>
> This time I asked a vector to be created with<br>
the stats and used this<br>
> to extract training polygons in QGIS and<br>
imported the training map in<br>
> GRASS. I had to do some interventions regarding<br>
the column names to<br>
> make sure they are the same except for the class.<br>
> I still get an error, and the only thing I could<br>
trace is the fact<br>
> that values are missing in some rows for both<br>
vectors. I am not sure<br>
> if I should correct this/ retry it all.<br>
<br>
I haven't seen this before, so yes, please try to<br>
eliminate the rows<br>
with missing values. How did you get the feature<br>
variables and how come<br>
there are missing values ?<br>
<br>
I don't have the time to test this right now, so I<br>
prefer not to commit<br>
as is, but you could try to edit your copy of<br>
v.class.mlR to add the<br>
four lines marked with a plus:<br>
<br>
r_file.write('features <- read.csv("%s",<br>
sep="%s", header=TRUE,<br>
row.names=1)' % (feature_vars, separator))<br>
r_file.write("\n")<br>
+ r_file.write("features <- na.omit(features)")<br>
+ r_file.write("\n")<br>
r_file.write('training <- read.csv("%s",<br>
sep="%s", header=TRUE,<br>
row.names=1)' % (training_vars, separator))<br>
r_file.write("\n")<br>
+ r_file.write("training <- na.omit(training)")<br>
+ r_file.write("\n")<br>
<br>
This would eliminate all lines that have at least<br>
one missing value.<br>
<br>
Another option would be for you to send me the<br>
data (segments<br>
and training) privately, so that I can test.<br>
<br>
Moritz<br>
<br>
<br>
><br>
> This is the command output:<br>
><br>
> (Fri Jun 08 15:48:28 2018)<br>
><br>
> v.class.mlR -i --overwrite<br>
><br>
segments_map=Segments_vector_<wbr>Stats_Ben_test@haarlooj_Ben_Te<wbr>st<br>
> training_map=Training_Ben5@haa<wbr>rlooj_Ben_Test<br>
><br>
raster_segments_map=best5_my<wbr>region1_at_haarlooj_Ben_Test_<wbr>rank1@haarlooj_Ben_Test<br>
> train_class_column=Ecosystem<br>
output_class_column=vote<br>
> output_prob_column=prob<br>
classifiers=svmRadial,rf,C5.0 folds=5<br>
> partitions=10 tunelength=10<br>
weighting_modes=smv,qbwwv<br>
> weighting_metric=accuracy<br>
><br>
classification_results=C:\Us<wbr>ers\haarlooj\Documents\CELOS\<wbr>v.class.mIRR_optional_output\<wbr>Ben_test_Classifier-results<br>
><br>
accuracy_file=C:\Users\haarl<wbr>ooj\Documents\CELOS\v.class.<wbr>mIRR_optional_output\Ben_test_<wbr>Classifier-accuracy<br>
><br>
model_details=C:\Users\haarl<wbr>ooj\Documents\CELOS\v.class.<wbr>mIRR_optional_output\Ben_test_<wbr>Classifier-module-runs<br>
><br>
bw_plot_file=C:\Users\haarlo<wbr>oj\Documents\CELOS\v.class.<wbr>mIRR_optional_output\Ben_test_<wbr>Classifier-performance<br>
><br>
r_script_file=C:\Users\haarl<wbr>ooj\Documents\CELOS\v.class.<wbr>mIRR_optional_output\Ben_test_<wbr>R_script<br>
> processes=3 Running R now. Following output is<br>
R output.<br>
> During startup - Warning messages:<br>
> 1: Setting LC_CTYPE=en_US.cp1252 failed<br>
> 2: Setting LC_COLLATE=en_US.cp1252 failed<br>
> 3: Setting LC_TIME=en_US.cp1252 failed<br>
> 4: Setting LC_MONETARY=en_US.cp1252 failed<br>
> Loading required package: caret<br>
> Loading required package: lattice<br>
> Loading required package: ggplot2<br>
> Loading required package: foreach<br>
> Loading required package: iterators<br>
> Loading required package: parallel<br>
> During startup - Warning messages:<br>
> 1: Setting LC_CTYPE=en_US.cp1252 failed<br>
> 2: Setting LC_COLLATE=en_US.cp1252 failed<br>
> 3: Setting LC_TIME=en_US.cp1252 failed<br>
> 4: Setting LC_MONETARY=en_US.cp1252 failed<br>
> During startup - Warning messages:<br>
> 1: Setting LC_CTYPE=en_US.cp1252 failed<br>
> 2: Setting LC_COLLATE=en_US.cp1252 failed<br>
> 3: Setting LC_TIME=en_US.cp1252 failed<br>
> 4: Setting LC_MONETARY=en_US.cp1252 failed<br>
> During startup - Warning messages:<br>
> 1: Setting LC_CTYPE=en_US.cp1252 failed<br>
> 2: Setting LC_COLLATE=en_US.cp1252 failed<br>
> 3: Setting LC_TIME=en_US.cp1252 failed<br>
> 4: Setting LC_MONETARY=en_US.cp1252 failed<br>
> Warning message:<br>
> In nominalTrainWorkflow(x = x, y = y, wts =<br>
weights, info =<br>
> trainInfo, : There were missing values in<br>
resampled performance<br>
> measures. Error in `$<-.data.frame`(`*tmp*`,<br>
vote_qbwwv, value =<br>
> numeric(0)) : replacement has 0 rows, data has 1965<br>
> Calls: $<- -> $<-.data.frame<br>
> Execution halted<br>
> ERROR: There was an error in the execution of<br>
the R script.<br>
> Please check the R output.<br>
> (Fri Jun 08 15:49:32 2018) Command finished (1<br>
min 4 sec)<br>
><br>
><br>
><br>
> Best,<br>
> Jamille<br>
><br>
><br>
><br>
><br>
> On Thu, Jun 7, 2018 at 11:09 AM, Jamille Haarloo<br>
> <<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><br>
<mailto:<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><wbr>><br>
<mailto:<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><br>
<mailto:<a href="mailto:j.r.haarloo@gmail.com" target="_blank">j.r.haarloo@gmail.com</a><wbr>>>> wrote:<br>
><br>
> > Hello Moritz,<br>
> ><br>
> > No worries. Thankful these modules are made<br>
available for newbies<br>
> > in RS like me and also happy these<br>
interactions are possible for<br>
> > learning. Hope to get back soon after some<br>
adjustments.<br>
> ><br>
> > Best,<br>
> > Jamille<br>
> ><br>
> > On Thu, Jun 7, 2018 at 10:44 AM, Moritz Lennert <<br>
> > <a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldonline.be</a><br>
<mailto:<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldon<wbr>line.be</a>><br>
<mailto:<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldon<wbr>line.be</a><br>
<mailto:<a href="mailto:mlennert@club.worldonline.be" target="_blank">mlennert@club.worldon<wbr>line.be</a>>>> wrote:<br>
> ><br>
> >> Thanks<br>
> >><br>
> >> On 07/06/18 15:17, Jamille Haarloo wrote:<br>
> >><br>
> >>> The first 20+ lines of Stats_Training_Ben_test:<br>
> >>><br>
> >>><br>
cat,area,perimeter,compact_ci<wbr>rcle,compact_square,fd,WV_Bena<wbr>t<br>
> >>><br>
imofo_1_min,WV_Benatimofo_1_m<wbr>ax,WV_Benatimofo_1_range,WV_Be<wbr>n<br>
> >>><br>
atimofo_1_mean,WV_Benatimofo_<wbr>1_stddev,WV_Benatimofo_1_varia<br>
> >>><br>
nce,WV_Benatimofo_1_coeff_<wbr>var,WV_Benatimofo_1_sum,WV_<br>
> >>><br>
Benatimofo_1_first_quart,WV_B<wbr>enatimofo_1_median,WV_Benatim<br>
> >>><br>
ofo_1_third_quart,WV_Benatimo<wbr>fo_2_min,WV_Benatimofo_2_max,<br>
> >>><br>
WV_Benatimofo_2_range,WV_Bena<wbr>timofo_2_mean,WV_Benatimofo_2_<br>
> >>><br>
stddev,WV_Benatimofo_2_varian<wbr>ce,WV_Benatimofo_2_coeff_var,<br>
> >>><br>
WV_Benatimofo_2_sum,WV_Benati<wbr>mofo_2_first_quart,WV_<br>
> >>><br>
Benatimofo_2_median,WV_Benati<wbr>mofo_2_third_quart,WV_Benatimo<wbr>f<br>
> >>><br>
o_3_min,WV_Benatimofo_3_max,W<wbr>V_Benatimofo_3_range,WV_Benat<br>
> >>><br>
imofo_3_mean,WV_Benatimofo_3_<wbr>stddev,WV_Benatimofo_3_varianc<br>
> >>><br>
e,WV_Benatimofo_3_coeff_var,W<wbr>V_Benatimofo_3_sum,WV_<br>
> >>><br>
Benatimofo_3_first_quart,WV_B<wbr>enatimofo_3_median,WV_Benatim<br>
> >>><br>
ofo_3_third_quart,WV_Benatimo<wbr>fo_4_min,WV_Benatimofo_4_max,<br>
> >>><br>
WV_Benatimofo_4_range,WV_Bena<wbr>timofo_4_mean,WV_Benatimofo_4_<br>
> >>><br>
stddev,WV_Benatimofo_4_varian<wbr>ce,WV_Benatimofo_4_coeff_var,<br>
> >>><br>
WV_Benatimofo_4_sum,WV_Benati<wbr>mofo_4_first_quart,WV_<br>
> >>> Benatimofo_4_median,WV_Benatim<wbr>ofo_4_third_quart<br>
> >>><br>
1144,3832.000000,1256.000000,<wbr>5.723635,0.197144,1.729624,13,<wbr>7<br>
> >>><br>
6,63,46.4097077244259,9.98454<wbr>911351384,99.69122100017,21.<wbr>513<br>
> >>><br>
9237092391,177842,40,47,53,<wbr>40,138,98,90.2687891440501,15.<wbr>250<br>
> >>><br>
0825418009,232.565017531741,1<wbr>6.8940812061464,345910,81,92,<br>
> >>><br>
100,15,61,46,40.<wbr>8582985386221,7.<wbr>82663897784868,61.2562776895<br>
> >>><br>
802,19.1555675536767,156569,3<wbr>6,42,47,28,124,96,68.42536534<br>
> >>><br>
44676,13.5774536655369,184.3<wbr>47248039801,19.8427200164517,2<wbr>62206,59,68,77<br>
> >>><br>
1145,12092.000000,2282.<wbr>000000,5.854120,0.192750,1.<wbr>645226,13,<br>
> >>><br>
94,81,51.386288455177,10.5294<wbr>376761475,110.869057775874,20.<wbr>4<br>
> >>><br>
907534532914,621363,45,52,59,<wbr>21,220,199,114.230731061859,23<wbr>.<br>
> >>><br>
3590328249442,545.64441451682<wbr>2,20.4489917973953,1381278,<wbr>101,<br>
> >>><br>
114,128,7,76,69,46.4219318557<wbr>724,8.42747122371732,71.<br>
> >>><br>
0222712265835,18.154072626491<wbr>5,561334,42,48,52,17,198,181,<br>
> >>><br>
97.2732385047966,22.492313569<wbr>247,505.904169697333,23.<br>
> >>> 1228176577445,1176228,84,97,11<wbr>0<br>
> >>><br>
> >>> [...]<br>
> >><br>
> >> ---------------------<br>
> >>> All the lines of the output of v.db.select<br>
> >>> Training_Ben2@haarlooj_Ben_Tes t:<br>
> >>><br>
> >>> cat|id|Type|code<br>
> >>> 1|4|B29|18<br>
> >>> 2|5|B31|19<br>
> >>> 3|3|B28|17<br>
> >>><br>
> >><br>
> >><br>
> >> Again a lack of clear documentation on my<br>
side: both the training<br>
> >> and the segment info should contain the same<br>
attributes, with only<br>
> >> additional one column ('code' in your case)<br>
present in the<br>
> >> training data.<br>
> >><br>
> >> It should be possible to do this<br>
differently, i.e. provide the<br>
> >> module with the features of all segments,<br>
and only the id/cat of<br>
> >> each training segment with the relevant<br>
class and have the module<br>
> >> merge the two, but this is not implemented, yet.<br>
> >><br>
> >> I also just notice that you have the word<br>
'Training' in both<br>
names.<br>
> >><br>
> >> The segment_file/segment_map contains the<br>
info (cat + all feature<br>
> >> variables) of all segments you wish to<br>
classify, either in the<br>
> >> form of a csv file or in the form of a<br>
vector map with the info in<br>
> >> the attribute table.<br>
> >><br>
> >> The training_file/training_map contains the<br>
info (cat + all<br>
feature<br>
> >> variables + class) of the training data.<br>
Often this is an extract<br>
> >> of the former, but not necessarily.<br>
> >><br>
> >> All columns in the training file have to be<br>
present in the segment<br>
> >> file, except for the class column (your 'code').<br>
> >><br>
> >> Sorry for the lack of docs. This module has<br>
mostly been used<br>
> >> internally here and so we are not always<br>
aware of the unclear and<br>
> >> missing parts. Having your feedback has been<br>
very useful !<br>
> >><br>
> >> Moritz<br>
> >><br>
> >><br>
> ><br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
</blockquote>
<br>
<br></div></div><span class="">
______________________________<wbr>_________________<br>
grass-user mailing list<br>
<a href="mailto:grass-user@lists.osgeo.org" target="_blank">grass-user@lists.osgeo.org</a><br>
<a href="https://lists.osgeo.org/mailman/listinfo/grass-user" rel="noreferrer" target="_blank">https://lists.osgeo.org/mailma<wbr>n/listinfo/grass-user</a><br>
<br>
</span></blockquote>
<br>
<br>
</blockquote></div><br></div>