Extracting predictive results for each loop in a K-fold cross validation or getting the standard deviation for each fold

KNIME Analytics Platform
1

Dear all

I’ve submitted an article where a reviewer is asking for standard deviation for all scores given (that is: accuracy, f1, cohen’s kappa and AUC) but I have no idea how to provide this with the knime analytics framework.
Ideally I would be able to calculate them in KNIME but I can also extract predictive scores all over and run them on a pyhton script. How would I in that case do that?

Did anyone find a solution for such a request?
I’m running the workflow in a classic loop with a 5-fold cross validation

Best, Chris

2

have you already tried saving metrics within the cv loop and store/append them in variables / tables “before” the cv loop closes (Nodes like scorer or ROC Cuve,… should be able to return results)
br

3

GBT.knwf (26.3 KB)

Dear Daniel.

Please find attached my workflow

I’ve tried with several notes and I just can’t get it to function. Have you done it before?

4

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.