Dear all,

I am trying to use the command that STATA has for DML (Cross-fit partialing-out lasso linear regression; xporegress) but I have some questions for which I could not find a solution:
1. I know that usually with Lasso methods we have to standardize the variables. I am wondering, is the command doing this automatically ? Moreover, as DML is not about predictions but inference, should the variables of interest also be standardized?
2. About this standardization procedure, I know there is a whole debate about whether categorical variables should be standardized. I have a few as well as controls and I am currently just transforming them to be considered as dummy variables. (i. command). Do you think that's enough ?
3. Lastly, I am also considering applying a log-scale to some of my continuous variables (e.g average income and expenditures). Do you think this is feasible after standardizing the variables?
4. Is there a way to create automatically interactions between variables as controls ?
To put it all in context, I am trying to estimate the effect of two variables (one categorical/dummy and one a continuous one) while taking into account the effect of a bunch of controls.
If you have any ideas about these questions or any general comments, please let me know.
Many thanks in advance.