Selected variable list using lasso

[Chinmay Sharma]

Hi All,

I have data which resembles the following:



* Example generated by -dataex-. To install: ssc install dataex
clear
input double(y x1 x2 year)
2 2 2 1991
3 3 3 1990
1 4 4 1990
4 1 1 1991
1 2 2 1990
4 3 3 1990
end
[/CODE]

Here, I wish to run a lasso regression of y on x1, x2 and year dummies, in order to select relevant regressors (between x1 and x2 in this example). In order to do this, I run:

lasso2 y x1 x2  (year) , lic(aicc)  long

Here, I choose the model using AICC, and I insert parentheses in order to specify that year dummies should be included, regardless of whether or not they are assigned a 0 coefficient using lasso. In doing so, however, the year dummies which are not significant, have still be dropped. I have two questions.

1. How can I force the model to include all year dummies in this case, and only select between x1 and x2?
2. How can I store the variable list which is selected in a local macro for use at a later point?

Many thanks for your consideration.

CS

[Jared Greathouse]

You can't. And I don't think it's possible aside from modifying the command

[Chinmay Sharma]

You can't. And I don't think it's possible aside from modifying the command

Thank you. Although I understand I cannot force some variables to be included no matter what, I must be able to retrieve the selected variable list, no?

[Jared Greathouse]

The way you CAN do it is either by using the predict postestimation command (which will show you, but can't save it), or.......... you can do what Fernando says here

[Chinmay Sharma]

Thank you so much!! Very helpful.