lassologit: Stata package for logistic lasso regression

[Achim Ahrens]

Following up on lassopack, Chris Hansen, Mark Schaffer and myself have developed a package for logistic lasso regression which can be used for prediction/classification tasks with binary outcomes. The package lassologit is not yet on SSC, but you can install it from our website:

net install lassologit, from("https://raw.githubusercontent.com/statalasso/lassologit/master/lassologit_v01/")

A demonstration of how lassologit works can be found in the help file ("help lassologit") and you can also find futher information on our website: https://statalasso.github.io/

Feedback and comments are much appreciated.

[Lydia Palumbo]

Dear Achim Ahrens,

I run this command on my model as I would like to select a more parsimonious model than the one that I have.

However, I have a couple of questions/issues about it.

1) It is still quite unclear to me how to interpret the output.
2) Why does lassologit allow me to run a regression with factor variables and, then, when I run it, it does not allow me to plot the coefficients path over lambda?
3) I guess that there is not possibility of running this command by using svy data...

Thank you very much.
Lydia