[An Dao]

Dear Clyde,

Thank you so much for your advice that made my command get through.
I am also learning how to screen out relevant variables for predicting an outcome. My outcome is dental caries which is the fiveyodmfs variable in the example data set in #12. This fiveyodmfs variable has a zero inflated distribution. I would like to apply lasso model. Unfortunately, it seems that lasso only applies with five options: logit, probit, reg, cox, and poisson. I wonder whether it is possilble to apply to zero inflated data? please give your advice. Your supports are highly appreciated.

[Ilya Bolotov]

If I constructed the dataset, I'd adjust the using data, based on specified conditions, and run

merge 1:1 [varlist] using {path to using data}, update

for the master ones.

Unless you are writing an ADO file, you can make any changes in both datasets before the merge.
Overall, append simply adds observations at the end of the file, joinby creates pairwise combinations based on a varlist, so merge is a better choice here.

[Clyde Schechter]

Re #16: You are raising a question unrelated to anything previous in this thread. (Actually your other posts in this thread are also off the original topic, but I didn't notice that or I would have said something earlier.) Please post your question as a New Topic. It is important to keep threads on topic so that others searching for answers to their questions can find what they are looking for, and also not waste their time reading unrelated material.

When you do repost in a new thread, it is best not to address it to me or any other specific person.. There are other Forum members who are more knowledgeable about this question than I am, and you only hurt your chances of getting good information by discouraging them from answering you.

[An Dao]

Yes, thank you for your advice, will do that. Appreciated your consultation.