cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Register for our Discovery Summit 2024 conference, Oct. 21-24, where you’ll learn, connect, and be inspired.
Choose Language Hide Translation Bar
gemmahodgson
Level III

How do I get the value of the R-Sq, L1, L2 and lambda out of Elastic Net Regression platform so I can compare models?

I am comparing different parameter settings for some elastic net regression models on a smallish data set to demonstrate at what point datasets are too small for regression methods and want to compare between JMP, SAS, R and Python. Although I can set the alpha, I can't see how to directly output the lambda, L1 and L2 values that are used in the chosen models and the resulting R-Sq for the model. Is there a secret right click somewhere I'm missing as I would've thought these are pretty straightforward things that people want to see? And the other software seem to give these in varying ways.

I am using JMP Po 18 and using the Leave one out cross validation method - which also seems to mean no model comparison summary statistics are generated (they are if I choose AICc or BIC as the CV method)...is that right?

Thanks in advance.

2 REPLIES 2

Re: How do I get the value of the R-Sq, L1, L2 and lambda out of Elastic Net Regression platform so I can compare models?

You can interactively extract these values by right-clicking on the column of numbers and selecting Make Into Data Table. You can script this action or a more specific extraction.

gemmahodgson
Level III

Re: How do I get the value of the R-Sq, L1, L2 and lambda out of Elastic Net Regression platform so I can compare models?

Thanks for this Mark, but I can't even find the values in the output to be able to do that. A reply from a JMP technician said "they don't surface these values", which I find a bit hard to understand as surely one wants to be able to see the lambda, L1 and L2 at the very least. Unless they are called by a different name in the JMP output?

They also said you don't get an R-Sq if you use K-Fold or LOO as validation - but no explanation as to why not - any ideas? I'm pretty sure SAS gives these directly so I'm not sure why JMP is not doing it?