topic Lasso Logistic Regression in Statistical Procedures

Lasso Logistic Regression

sas_sad — Thu, 11 Feb 2021 01:46:39 GMT

Hi Folks,

I am running a logistic model where my outcome is 0/1 and independent variables are either 0/1 or quartiles. The age variable (20 to 54) is highly collinear with education and race (VIF is 12). However, I cannot drop age and I tried recategorizing it, but the VIF was still above 10. So, I started reading that it can be taken care of by doing a ridge or lasso regression. I have searched extensively, but I did not get any information on how to implement it in logistic regression along with interpretation.

Thanks

Re: Lasso Logistic Regression

Reeza — Thu, 11 Feb 2021 02:47:25 GMT

Example 3 here: https://support.sas.com/kb/60/240.html

Re: Lasso Logistic Regression

sas_sad — Thu, 11 Feb 2021 03:46:27 GMT

I did this, but the VIF was still around 12. I categorized age into smaller groups, but both had vid above 10.

Re: Lasso Logistic Regression

Reeza — Thu, 11 Feb 2021 16:29:53 GMT

What exactly do you mean when you say you "did this"? What is "this"?
Because nothing in the link above is about changing the categories into smaller groups.

Re: Lasso Logistic Regression

sas_sad — Thu, 11 Feb 2021 17:38:58 GMT

I am sorry, I meant this: