1
$\begingroup$

In R, when we use glmnet package. We use cv function as cross validation in finding the value of lambda. In the package, we will find two options in the bottom, lambda.min and lambda.1se. If I use Lasso selection, which lambda should I pick in Multinomial Logistics Regression using Lasso?

Some recommended in using lambda.1se as it is simpler and comparable to the best model. Yet, I cannot find one reliable citation.

Please help, thank you.

$\endgroup$

marked as duplicate by Sycorax, kjetil b halvorsen, mdewey, Peter Flom Oct 3 '17 at 11:15

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • 2
    $\begingroup$ If you would read the glmnet manual I believe you will find citations. $\endgroup$ – whuber Oct 2 '17 at 15:05
-1
$\begingroup$

Reasoning is to choose the most parsimonious model within 1 SE from the best model the optimizer found. In my experience, this rule of thumb does not always work. But at least it gives you some leeway to investigate anything in between.

Why is lambda plus 1 standard error a recommended value for lambda in an elastic net regression?

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.