Error: ssc install: "dsregress" not found at SSC

[Scott Rick]

Hi there,

I want to need a method for inference with variable selection.
To do so, I tried the dsregress command to apply lasso variable selection and regression, modelled in the following notations

 
  dsregress Y X1, controls(X2-X50) sel(cv)

However, stata does not appear to identify this command and neither can I seem to install it.

. dsregress con_avg7day i.Stepone, controls( rate_pneumoniaml_death2018 rate_pneumoniafem_case20
> 18 rate_pneumoniafem_death2018 rate_pneumoniatot_case2018 rate_pneumoniatot_death2018 rate_tb2
> 018 rate_influenzacase_2018 rate_influenzadeath_2018 rate_hepcases_2018 rate_hepdeaths_2018 ra
> te_diabetes2018) sel(cv)
command dsregress is unrecognized
r(199);

. ssc install dsregress
ssc install: "dsregress" not found at SSC, type search dsregress
(To find all packages at SSC that start with d, type ssc describe d)
r(601);

Has this command been deleted and is there an alternate one? If not, what can be done here?

[Nick Cox]

Indeed. The code for dsregress is not community-contributed and also not at SSC.

dsregress is part of official Stata and was first added in Stata 16.

It seems that you have a version of Stata not supporting dsregress and if so you must upgrade to be able to use it. See also the long-standing request at https://www.statalist.org/forums/help#version to tell us about use of a version before the most recent.

[Scott Rick]

Thanks Nick Cox. I'm using Stata 15.1, and have updated it. However, given the version that first added dsregress, I don't have access to it. As a student, I also cannot currently afford the update.

Is there any other command that does what dsregress does for Stata versions before version 16?

[Nick Cox]

I'd look at

SJ-20-1 st0594 . . Model selection and prediction with regularized regression
. . . . . . . . . . . . . A. Ahrens, C. B. Hansen, and M. E. Schaffer
(help cvlassologit, cvlasso, lasso2, lassologit, lassopack,
rlassologit, rlasso if installed)
Q1/20 SJ 20(1):176--235
provides a suite of programs for regularized regression:
lasso, square-root lasso, elastic net, ridge regression,
adaptive lasso, and postestimation ordinary least squares


to see if anything there helps.

[Scott Rick]

Thanks Nick Cox