Potential Bug With Forest-based and Boosted Classification And Regression Tool (Spatial Statistics) in ArcGIS Pro 3.2.1
Just posting this in case others have the same issue.
Issue:
When trying to "Predict to Raster" using the Forest-based and Boosted Classification And Regression tool, as soon as the message for "Cell Size" is printed, the tool ends as "Completed" without creating the model or predicting to a raster.
Settings:
- Model Type = Forest-based
- Input Training Features = Polygon
- Treat Variable as Categorical= Checked
- Only using explanatory training rasters
- Convert Polygons to Raster Resolution for Training = checked
- Polygon and rasters are all in the same projection
- Match Explanatory Rasters are all matched correctly
Comments:
Everything works as intended if "Train Only" is selected. After switching to "Predict to Raster" the tool fails as mentioned above. If I uncheck "Convert Polygons to Raster Resolution for Training", the tool will predict to raster as expected.
While using "Train Only", I can convert my training polygons to points using the "Output Trained Features" setting. If I use these points as the "Input Training Features", the tool will predict to raster without any issues (however, it should be noted that for some reason out-of-bag errors are worse when using these points vs the original polygons, which seems odd if the "Convert Polygons to Raster Resolution for Training" setting works as described).
Tool Settings
Tool Messages
Note: I get the above warnings even when the tool runs successfully.
