Lasso postestimation OLS

[Cecily Josten]

Hi! I am trying to run a Lasso regression for variable selection. Specifically I would like Stata to return the relevant coefficients from the Lasso approach and then run an OLS regression with those selected variables to get coefficients with standard errors. I am puzzled, however, which commands will get me there. I have been using "lasso linear" and then "lassocoef, display(coef, postselection)". I then read, however, to get standard errors I would need to run dsregress but this does not give me the coefficients for all my controls as I would like them to have. If you could explain that would be really helpful! Thank you!

[Andrew Musau]

The selected variables are stored in the macro -e(post_sel_vars)-. As your model is linear, you can use regress on the post-selection variables.

webuse cattaneo2
lasso linear bweight c.mage##c.mage c.fage##c.fage ///
c.mage#c.fage c.fedu##c.medu i.(mmarried mhisp ///
fhisp foreign alcohol msmoke fbaby prenatal1)

regress `e(post_sel_vars)'

Res.:

. regress `e(post_sel_vars)'

      Source |       SS           df       MS      Number of obs   =     4,642
-------------+----------------------------------   F(10, 4631)     =     29.44
       Model |    92938669        10   9293866.9   Prob > F        =    0.0000
    Residual |  1.4619e+09     4,631  315686.799   R-squared       =    0.0598
-------------+----------------------------------   Adj R-squared   =    0.0577
       Total |  1.5549e+09     4,641  335032.156   Root MSE        =    561.86

------------------------------------------------------------------------------
     bweight |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        mage |   .4884131   1.840359     0.27   0.791    -3.119568    4.096394
        fedu |   2.238029   2.871845     0.78   0.436    -3.392155    7.868213
        medu |   4.848398   4.206963     1.15   0.249    -3.399253    13.09605
    mmarried |   150.5516    22.2432     6.77   0.000     106.9443    194.1588
       mhisp |   66.69611    49.8213     1.34   0.181    -30.97737    164.3696
     foreign |  -85.81082   39.73678    -2.16   0.031    -163.7138   -7.907803
     alcohol |  -57.24867   47.53726    -1.20   0.229    -150.4443      35.947
      msmoke |  -94.66951   9.680001    -9.78   0.000    -113.6469   -75.69209
       fbaby |  -58.15499    17.9433    -3.24   0.001     -93.3324   -22.97759
   prenatal1 |   48.49205   22.54524     2.15   0.032     4.292634    92.69146
       _cons |    3182.93   55.17699    57.69   0.000     3074.757    3291.104
------------------------------------------------------------------------------