1

I'm trying to run a Random Forest Classification and Regression in ArcGIS Pro 2.3.2. I've already run this analysis using Bayesian logistic regression in R, and want to compare the results from RF.

I have options set to Prediction Type = Raster, am using a polygon layer as the Training Features, selected the Variable to Predict and checked Categorical (it's a 0/1 presence/absence variable), input Training Rasters and checked the ones that are categorical, and specified an Output Prediction Surface (see image below).

enter image description here

The process makes it past "Reading Data" and "Getting Centroids", but then immediately fails with the following error:

Traceback (most recent call last):

File "<string>", line 10175, in execute

File "C:\Program Files\ArcGIS\Pro\Resources\ArcToolbox\Scripts\SSForest.py", line 6638, in execute created = rasterWorkFlow(*infoRasterArg)

File "C:\Program Files\ArcGIS\Pro\Resources\ArcToolbox\Scripts\SSForest.py", line 6441, in rasterWorkFlow checkBalance = not hpar.balanceTree)

File "C:\Program Files\ArcGIS\Pro\Resources\ArcToolbox\Scripts\SSForest.py", line 216, in loadDataRasterForTraining self.createVariablesFromRasterList(infoRaster, dataX)

File "C:\Program Files\ArcGIS\Pro\Resources\ArcToolbox\Scripts\SSForest.py", line 284, in createVariablesFromRasterList errors, info = self.variableTest(name, dataBlock, userType, id)

File "C:\Program Files\ArcGIS\Pro\Resources\ArcToolbox\Scripts\SSForest.py", line 318, in variableTest info = field.info

AttributeError: 'NoneType' object has no attribute 'info'

Failed to execute (Forest).

Obviously, the SSForest.py script is failing at specific lines, but I don't know what the issue is. All the lines in the error message seem to relate to the loadDataRasterForTraining (line 203) and loadDataRasterForPredicting(line 252) functions in the script. Any ideas what might be causing the error?

0

I managed to figure out the problem after digging through the code for SSForest.py and trying different combinations of input data in the geoprocessing menu.

The Variable to Predict from your Training Features must be a text string. Using a field with categorical numeric codes causes the tool to fail, even after checking Treat Variable as Categorical.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.