xuyiqing / fect

fixed effect counterfactual estimators
Other
53 stars 20 forks source link

Different optimal lambda values with and without specifying the range of lambda #26

Closed Yanxia-Yu closed 1 year ago

Yanxia-Yu commented 2 years ago

Hi Yiqing,

I did the following things and found that I get slightly different estimates and standard errors and very different lambda values when I specified or did not specify the range of lambda.

My experiment consists of two steps:

1) Do not specify the range of lambda and let the computer program to decide; 2) add the optimal lambda value generated in step 1 to the range of lambda and rerun the code.

Step 1 and step 2 return different lambda values. I'm wondering whether we should specify the range of lambda. What's your thoughts on that? And what does "lambda.norm" in the output mean?

Thank you very much for your help. Attached are my data and code.

Yanxia Data & Code.zip