Print each command in loop

[Steven Beaker]

Does anyone have suggestions on how to make a more descriptive log file within a loop. For example, if you're running multiple regressions and you want to see the command used to generate the output of each regression, is there a way to log each iteration of the loop? Here's a more detailed example:

log using "example.txt", name(example) text replace

sysuse auto.dta

* Not that helpful in a log file
forvalues i = 0/1 {
    regress mpg price if foreign == `i'
}

* More helpful in a log file
regress mpg price if foreign == 0
regress mpg price if foreign == 1

log close example

Here's the output of this script:

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
      name:  example
       log:  example.txt
  log type:  text
 opened on:  23 Dec 2022, 15:19:05

. 
. sysuse auto.dta
(1978 automobile data)

. 
. * Not that helpful in a log file
. forvalues i = 0/1 {
  2.      regress mpg price if foreign == `i' 
  3. }

      Source |       SS           df       MS      Number of obs   =        52
-------------+----------------------------------   F(1, 50)        =     17.05
       Model |  291.772809         1  291.772809   Prob > F        =    0.0001
    Residual |  855.669499        50    17.11339   R-squared       =    0.2543
-------------+----------------------------------   Adj R-squared   =    0.2394
       Total |  1147.44231        51  22.4988688   Root MSE        =    4.1368

------------------------------------------------------------------------------
         mpg | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       price |  -.0007723    .000187    -4.13   0.000     -.001148   -.0003966
       _cons |   24.51661   1.272427    19.27   0.000     21.96086    27.07235
------------------------------------------------------------------------------

      Source |       SS           df       MS      Number of obs   =        22
-------------+----------------------------------   F(1, 20)        =     13.25
       Model |  365.808088         1  365.808088   Prob > F        =    0.0016
    Residual |  552.055549        20  27.6027774   R-squared       =    0.3985
-------------+----------------------------------   Adj R-squared   =    0.3685
       Total |  917.863636        21  43.7077922   Root MSE        =    5.2538

------------------------------------------------------------------------------
         mpg | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       price |  -.0015918   .0004373    -3.64   0.002     -.002504   -.0006797
       _cons |    34.9361   3.008144    11.61   0.000     28.66122    41.21097
------------------------------------------------------------------------------

. 
. * More helpful in a log file
. regress mpg price if foreign == 0

      Source |       SS           df       MS      Number of obs   =        52
-------------+----------------------------------   F(1, 50)        =     17.05
       Model |  291.772809         1  291.772809   Prob > F        =    0.0001
    Residual |  855.669499        50    17.11339   R-squared       =    0.2543
-------------+----------------------------------   Adj R-squared   =    0.2394
       Total |  1147.44231        51  22.4988688   Root MSE        =    4.1368

------------------------------------------------------------------------------
         mpg | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       price |  -.0007723    .000187    -4.13   0.000     -.001148   -.0003966
       _cons |   24.51661   1.272427    19.27   0.000     21.96086    27.07235
------------------------------------------------------------------------------

. regress mpg price if foreign == 1 

      Source |       SS           df       MS      Number of obs   =        22
-------------+----------------------------------   F(1, 20)        =     13.25
       Model |  365.808088         1  365.808088   Prob > F        =    0.0016
    Residual |  552.055549        20  27.6027774   R-squared       =    0.3985
-------------+----------------------------------   Adj R-squared   =    0.3685
       Total |  917.863636        21  43.7077922   Root MSE        =    5.2538

------------------------------------------------------------------------------
         mpg | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       price |  -.0015918   .0004373    -3.64   0.002     -.002504   -.0006797
       _cons |    34.9361   3.008144    11.61   0.000     28.66122    41.21097
------------------------------------------------------------------------------

. 
. log close example
      name:  example
       log:  example.txt
  log type:  text
 closed on:  23 Dec 2022, 15:19:05
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

[Joro Kolev]

You can use -display- to show whatever you find suitable, e.g.

sysuse auto.dta


forvalues i = 0/1 {
    display "regress mpg price if foreign == `i'"
    regress mpg price if foreign == `i'
}

[William Lisowski]

Building on Joro's answer, you may find this helpful.

sysuse auto.dta
capture program drop v
program define v
    macro shift
    display _newline `". `0'"'
    `0'
end
forvalues i = 0/1 {
    v regress mpg price if foreign == `i' 
}

. forvalues i = 0/1 {
  2.     v regress mpg price if foreign == `i' 
  3. }

. regress mpg price if foreign == 0

      Source |       SS           df       MS      Number of obs   =        52
-------------+----------------------------------   F(1, 50)        =     17.05
       Model |  291.772809         1  291.772809   Prob > F        =    0.0001
    Residual |  855.669499        50    17.11339   R-squared       =    0.2543
-------------+----------------------------------   Adj R-squared   =    0.2394
       Total |  1147.44231        51  22.4988688   Root MSE        =    4.1368

------------------------------------------------------------------------------
         mpg | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       price |  -.0007723    .000187    -4.13   0.000     -.001148   -.0003966
       _cons |   24.51661   1.272427    19.27   0.000     21.96086    27.07235
------------------------------------------------------------------------------

. regress mpg price if foreign == 1

      Source |       SS           df       MS      Number of obs   =        22
-------------+----------------------------------   F(1, 20)        =     13.25
       Model |  365.808088         1  365.808088   Prob > F        =    0.0016
    Residual |  552.055549        20  27.6027774   R-squared       =    0.3985
-------------+----------------------------------   Adj R-squared   =    0.3685
       Total |  917.863636        21  43.7077922   Root MSE        =    5.2538

------------------------------------------------------------------------------
         mpg | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       price |  -.0015918   .0004373    -3.64   0.002     -.002504   -.0006797
       _cons |    34.9361   3.008144    11.61   0.000     28.66122    41.21097
------------------------------------------------------------------------------

.

[Steven Beaker]

Thanks William and Joro! I should have mentioned I typically use display, but it becomes tedious with many commands that take place within a loop.

Joro's solution:

sysuse auto.dta


forvalues i = 0/1 {
    display "summarize mpg price if foreign == `i'"
    summarize mpg price if foreign == `i'
    display "regress mpg price if foreign == `i'"
    regress mpg price if foreign == `i'
}

William's clever solution saves lines, but requires prefixing each line with the new command.

forvalues i = 0/1 {
    v summarize mpg price if foreign == `i'
    v regress mpg price if foreign == `i'
}

William's may be the best solution for me, but I'm curious if there's a way to utilize "set trace" and "set tracedepth" so I don't have to prefix every command with another new command.

[Joro Kolev]

I do not think that -set trace- would help here, because the expansion is in "the other direction," but I might be wrong...

Here is another solution that does what you want, and we are not supposed to do such things because Stata Corp has discontinued the old looping construct -for-.

sysuse auto.dta

for numlist 0/1:    regress mpg price if foreign == X

results in the execution

. sysuse auto.dta
(1978 automobile data)

. 
. 
. for numlist 0/1:    regress mpg price if foreign == X

->  regress mpg price if foreign == 0

      Source |       SS           df       MS      Number of obs   =        52
-------------+----------------------------------   F(1, 50)        =     17.05
       Model |  291.772809         1  291.772809   Prob > F        =    0.0001
    Residual |  855.669499        50    17.11339   R-squared       =    0.2543
-------------+----------------------------------   Adj R-squared   =    0.2394
       Total |  1147.44231        51  22.4988688   Root MSE        =    4.1368

------------------------------------------------------------------------------
         mpg | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       price |  -.0007723    .000187    -4.13   0.000     -.001148   -.0003966
       _cons |   24.51661   1.272427    19.27   0.000     21.96086    27.07235
------------------------------------------------------------------------------

->  regress mpg price if foreign == 1

      Source |       SS           df       MS      Number of obs   =        22
-------------+----------------------------------   F(1, 20)        =     13.25
       Model |  365.808088         1  365.808088   Prob > F        =    0.0016
    Residual |  552.055549        20  27.6027774   R-squared       =    0.3985
-------------+----------------------------------   Adj R-squared   =    0.3685
       Total |  917.863636        21  43.7077922   Root MSE        =    5.2538

------------------------------------------------------------------------------
         mpg | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       price |  -.0015918   .0004373    -3.64   0.002     -.002504   -.0006797
       _cons |    34.9361   3.008144    11.61   0.000     28.66122    41.21097
------------------------------------------------------------------------------

[Steven Beaker]

Joro,

That's a neat trick! I'll have to think more carefully about my approach with loops. Thanks for entertaining my question! It's very helpful to know that there's not a straightforward built-in way to proceed for me. I appreciate it!

[Joro Kolev]

Joro,

That's a neat trick! I'll have to think more carefully about my approach with loops. Thanks for entertaining my question! It's very helpful to know that there's not a straightforward built-in way to proceed for me. I appreciate it!

You are welcome, Steven.

This sounds like a reasonable wish for Stata Corp, for them to offer that loops are executed noisily and somehow show the commands you are executing. The new loop constructs are in the core of Stata, the old -for- is an ado file.

. which for
C:\Program Files\Stata17\ado\base\f\for.ado
*! version 3.1.13  16feb2015

. which foreach
built-in command:  foreach

. which forvalues
built-in command:  forvalues

[Bjarte Aagnes]

prog define code // to be traced

sysuse auto.dta

forvalues i = 0/1 {
    
    regress mpg price if foreign == `i'
}

end 

set trace on
set tracedepth 1

log using "example.txt",   name(example) text replace

code // to be traced

log close example

[Bjarte Aagnes]

#8 need set traceexpand on