Neural Network Errors

[Riot]

Neural Network (118) Neural Network: Error: cannot allocate vector of size 261.3 Mb
Neural Network (118) Neural Network: Execution halted
Render (119) Data Field not found.
Neural Network (118) Neural Network: The R.exe exit code (4294967295) indicated an error.
Score (62) An empty or invalid spatial object was found. Consider filtering them out if you are using Spatial Statistics.
Score (62) Score: Error in getXVars(mod.obj$call) : The argument was not a call object.
Score (62) Score: Execution halted
Score (62) Score: The R.exe exit code (1) indicated an error.
Designer x64 Finished running in 1:02 hours with 7 errors and 55 field conversion errors and 5 warnings

Ran for an hour and I got this :?

Any idea? Is my database too big? Should I sample?

[RobertBl · Accepted solution]

Riot,

There are a couple of things I'd try here.

First off I see the message for the Score tool and the spatial object. Do you have a spatial object in your dataset and is it necessary for the analysis? You could remove it temporarily with a Select Tool and see how it runs.

Yes, I'd also suggest running it with a small sample size and seeing if it executes.

Those would be my first checks.

FYI  - Going a little further down the road it is sometimes helpful with the predictive tools (which are macros) to turn on the messages from the macros. That way you can open the macro and see the tools within the macro with errors.

- In your workflow click on the canvas and then click on the Runtime tab of the Workflow configuration and click the "Show all macro messages checkbox"

- Right click on the  Neural Network tool and choose the "Open Macro:..." choice

- When you run the workflow and see messages from the Macro note the tool name and number

- You can then switch to the Macro, right click on the canvas and choose "Find Tool" to find the tool with that name and number