Re: st: Foreach looping code for generating a Quantiles variable for each two digit sic code

[Nick Cox <njcoxstata@gmail.com>]

My typo, sorry

             egen resremRDq`y' = xtile(resremRD) if Year ==`y'

As for your question: If the data have a different structure or shape,
then the code would need to differ. What -reshape- calls "long" is in
general a better structure for panel data.

Nick

On Thu, Feb 16, 2012 at 10:19 PM, Jaime Laird <jmc981@gmail.com> wrote:
> Nick,
>
> Thank you for the reply! That was very helpful. Yes, I did omit an -if- condition from his original code. I also forgot to mention that my code I listed below was set up to work with the data in long format when I had a Year variable but I was hoping to be able to loop the command without switching to long.
>
> One question: Where does the `v' come in? What does it mean?
>
>>                               egen resremRDq`y' = xtile(resremRD) if Year ==`v' ,
>
>
>
>> However, why use separate variables for each year when you could put
>> the results in one variable?
>>
>>      egen resremRDq = xtile(resremRD), nquantiles(4) by(twodigit year)
>
> **This is exactly the end result I wanted! Thank you for this. I assume I will need to format long to get the year variable to use this also?
>
>
>
>
> On Feb 16, 2012, at 4:02 PM, Nick Cox wrote:
>
>> The trickiest part here is that years are represented inconsistently
>> but otherwise the loop is just a loop over years. Study of -help
>> numlist- points to a more direct solution:
>>
>> foreach y of num 95/99 2000/2010 {
>>      egen resremRDq`y' = xtile(resremRD) if Year ==`v' ,
>> nquantiles(4) by(twodigit)
>> }
>>
>> Naturally all you need to do is
>>
>> replace Year = Year + 1900 if Year < 100
>>
>> and then the loop could just be
>>
>> foreach y of num 1995/2010 {
>>      egen resremRDq`y' = xtile(resremRD) if Year ==`v' ,
>> nquantiles(4) by(twodigit)
>> }
>>
>> or
>>
>> forval y = 1995/2010 {
>>      egen resremRDq`y' = xtile(resremRD) if Year ==`v' ,
>> nquantiles(4) by(twodigit)
>> }
>>
>> However, why use separate variables for each year when you could put
>> the results in one variable?
>>
>>      egen resremRDq = xtile(resremRD), nquantiles(4) by(twodigit year)
>>
>> There is a basic tutorial on loops in
>>
>> SJ-2-2  pr0005  . . . . . .  Speaking Stata:  How to face lists with fortitude
>>        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J. Cox
>>        Q2/02   SJ 2(2):202--222                                 (no commands)
>>        demonstrates the usefulness of for, foreach, forvalues, and
>>        local macros for interactive (non programming) tasks
>>
>> The strange but to me pleasing fact that the work appeared in 2(2):
>> 202-222 (2002) was a pure accident, but too too droll not to mention.
>>
>> Back to your code: The loop
>>
>> foreach t of loc ts {
>>     xtile resremRDq`yr' = resremRD`yr', nquantiles(4)
>> }
>>
>> fails because you make no use of the values of -ts- within the loop.
>> Thus second time round the loop the -xtile- call is exactly the same
>> as first time round, and your code stops within each error. I guess
>> you omitted an -if- condition.
>>
>> Nick
>>
>> On Thu, Feb 16, 2012 at 9:14 PM, Jaime Laird <jmc981@gmail.com> wrote:
>>
>>> I have search the appropriate places and have not seen my particular question addressed so I hoping I will add something to the list with this question. Forgive me if I have missed something.
>>>
>>> What I need to do:
>>>
>>> I need to plot the Quartiles for each two-digit Siccode in my data and it also need to be done by year. I have successfully found the code to do this but I would like to better understand looping to streamline tasks such as these.
>>>
>>> The following code gives the results I need (I downloaded this SSC from within Stata, it is called egenmore):
>>>
>>> egen resremRDq96 = xtile (resremRD), by (twodigit), if Year ==96, nquantiles(4)
>>> egen resremRDq97 = xtile (resremRD), by (twodigit), if Year ==97, nquantiles(4)
>>> egen resremRDq98 = xtile (resremRD), by (twodigit), if Year ==98, nquantiles(4)
>>> egen resremRDq99 = xtile (resremRD), by (twodigit), if Year ==99, nquantiles(4)
>>> egen resremRDq2000 = xtile (resremRD), by (twodigit), if Year ==2000, nquantiles(4)
>>> egen resremRDq2001 = xtile (resremRD), by (twodigit), if Year ==2001, nquantiles(4)
>>> egen resremRDq2002 = xtile (resremRD), by (twodigit), if Year ==2002, nquantiles(4)
>>> egen resremRDq2003 = xtile (resremRD), by (twodigit), if Year ==2003, nquantiles(4)
>>> egen resremRDq2004 = xtile (resremRD), by (twodigit), if Year ==2004, nquantiles(4)
>>> egen resremRDq2005 = xtile (resremRD), by (twodigit), if Year ==2005, nquantiles(4)
>>> egen resremRDq2006 = xtile (resremRD), by (twodigit), if Year ==2006, nquantiles(4)
>>> egen resremRDq2007 = xtile (resremRD), by (twodigit), if Year ==2007, nquantiles(4)
>>> egen resremRDq2008 = xtile (resremRD), by (twodigit), if Year ==2008, nquantiles(4)
>>> egen resremRDq2009 = xtile (resremRD), by (twodigit), if Year ==2009, nquantiles(4)
>>>
>>> A member here gave me a code to help loop my regressions and I tried to adapt it to work with these quantiles but was unsuccessful. Here is the code I tried to make from his:
>>>
>>> forv y=1995/2009 {
>>> if `y'<2000 loc yr=substr("`y'",3,2)
>>> else loc yr `y'
>>> qui levelsof twodigit if !mi(rd`yr'), loc(ts)
>>> foreach t of loc ts {
>>> xtile resremRDq`yr' = resremRD`yr', nquantiles(4)
>>> }
>>> }
>>>
>>> The results only formed one new variable named "resremrdq95" and it did not assign the Quartiles by siccode (twodigit).
>>>
>>> It also returns this error:
>>>
>>> run "/var/folders/aL/aLs0EaEHFfCArViD4qvSeU+++TI/-Tmp-//SD07955.000000"
>>> resremRDq95 already defined
>>> r(110);
>>>
>>> I truly want to learn so please be patient with me. I am not simply looking for quick fixes. I want to be able to write my own codes effectively.
>>> In my mind, the above code should work because it is telling stata to look at each year in 'y', through the levels of twodigit. What am I missing?

*
*   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/