Predictions / Test & Score: include F-test of linear regression in evaluation results

[wvdvegte]

What's your use case?
What's your proposed solution?
To scientifically underpin results of Linear Regression, it would be nice to include the F-test for overall significance of the regression in the evaluation results. From the link above, it appears to be a simple calculation - it is the ratio between the MSE (already there) and the MSE based on the intercept-only version of the regression model (i.e., with all coefficients other than the intercept zeroed).
I know that Orange is not a statistical package, but a lot of stuff related to linear regression finds itself in the grey area between ML and statistics ...

Are there any alternative solutions?
I guess it would be possible to compute this with a Python script using the predictions and the coefficients as inputs.