Forest based classification and regression - error 110180: not enough variability

StephanKaegi

Hello

I have been using the "Forest-based classification and regression" tool and I am stuck on an error (not enough variability).

I run this tool using polygon dataset with 0 or 1 value (categorical) as training feature and raster files.

I cannot find the origin of this problem. What is meant by not enough Variability? 

Description

At least one field in the Explanatory Training Variables parameter contains records, and at least 95 percent of those records have the same value. More variability in the values is required.

Solution

Determine which field has many of the same values, and either remove the field or alter the values to provide more variability.

 

The print screen shows the histogram of the raster which has not enough variability - as we can see: not 95% of the records have the same value.

What can I do here?

 

 

Do you have any ideas how to solve it?

 

Thanks

 

 

 

 

Forestbased_printscreen.JPG

PatrickConnell

Had this same problem, knowing my training rasters were not similar enough to trigger this error. After checking out my Input Training Features, I saw that I had a few rows selected which was causing the variability error. Interesting to see the tool will work with selected features only! I'm sure you have it figured out by now, but posting in case anyone else encounters similar issue.