st: new version of -tryem- on SSC

["Feiveson, Alan H. (JSC-SK311)" <alan.h.feiveson@nasa.gov>]

Hi - Thanks to Kit Bauam, there is a slightly new verion of -tryem- on the SSC. -tryem- finds the best subset of k predictors by brute force (k specified by user) for a variety of estimation commands that can be run in  Stata. "Best" is defined as the subset of predictors that optimize the value of a statistic (such as log likelihood) specified by the user. The only difference between this and the last version is that now the "best" estimation can be replayed after running -tryem-.

Al Feiveson

Example:

. tryem y instantaneoushrbpm- v2unexqtvi,k(2) cmd(logit) stat(ll)
 --------------------------------------------
 
 Largest ll is     -781.17.  Best subset of size 2 is: 
 
 sd1 iirmssd_qt
 
 Variable numbers for best subset are:

  10  28
 --------------------------------------------

. logit

Logistic regression                               Number of obs   =       1919
                                                  LR chi2(2)      =    1089.00
                                                  Prob > chi2     =     0.0000
Log likelihood = -781.17222                       Pseudo R2       =     0.4107

------------------------------------------------------------------------------
           y |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         sd1 |  -.0777813   .0052755   -14.74   0.000     -.088121   -.0674415
  iirmssd_qt |   1.048285   .0630278    16.63   0.000     .9247527    1.171817
       _cons |  -1.035826   .1546785    -6.70   0.000    -1.338991   -.7326621
------------------------------------------------------------------------------
Note: 0 failures and 22 successes completely determined.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/