Re: st: Re: Replace current values matching certain condition using values from other observations?

[Chris Yang <bhpchg@gmail.com>]

Thank you Joseph.

As a related question, if I have many variables that need to be
replaced like this, is there a quick way to do them all at once
instead of doing it one-by-one for each variable. Perhaps I could use
a loop to loop through the variables, but if all the variable names
follow a pattern, e.g. var*, is there a shortcut?

I've tried: bysort group (seq): replace var*=var*[1] if seq > 1 for
the example given, but it gives an error saying: "var ambiguous
abbreviation".

On Sat, Jan 25, 2014 at 3:33 AM, Joseph Coveney <stajc2@gmail.com> wrote:
> Chris Yang wrote:
>
> I have a dataset that resembles the following structure:
>
> group seq var1 var2 var3
> 1       1     3      2     3
> 1       2     1      1     2
> 1       3     2      2     4
> 2       1     3      2     1
> 2       2     3      3     3
> 3       1     3      2     1
> ...
>
> Now, within each group, for all the observations whose seq > 1, I want
> to replace the values for var1 and var2 with those of the observation
> whose seq == 1. For example, for group 1 above, after the
> replacements, it would look like this:
>
> group seq var1 var2 var3
> 1       1     3     2      3
> 1       2     3     2      2
> 1       3     3     2      4
>
> If there is no seq > 1 in a given group, then no replacement is needed.
>
> Intuitively, it seems that I need a looping structure to go through
> all the observations one by one. And at each step i will check the seq
> variable. If it is greater than 1, then look up the values for var1
> and var2 from the observation *within the same group* whose seq == 1,
> and use them to update the current observation. The question is that
> how do I do such look-ups in a loop?
>
> As always, is there a better/more efficient way of doing it? Your
> thoughts and suggestions are appreciated.
>
> --------------------------------------------------------------------------------
>
> In Stata, you tend to avoid looping over data.  You can often take advantage of
> the fact that its data operations are "vectorized".
>
> Joseph Coveney
>
> . input byte(group seq var1 var2 var3)
>
>         group       seq      var1      var2      var3
>   1. 1       1     3      2     3
>   2. 1       2     1      1     2
>   3. 1       3     2      2     4
>   4. 2       1     3      2     1
>   5. 2       2     3      3     3
>   6. 3       1     3      2     1
>   7. end
>
> .
> . bysort group (seq): replace var1 = var1[1] if seq > 1
> (2 real changes made)
>
> . by group: replace var2 = var2[1] if seq > 1
> (2 real changes made)
>
> .
> . list, noobs sepby(group)
>
>   +----------------------------------+
>   | group   seq   var1   var2   var3 |
>   |----------------------------------|
>   |     1     1      3      2      3 |
>   |     1     2      3      2      2 |
>   |     1     3      3      2      4 |
>   |----------------------------------|
>   |     2     1      3      2      1 |
>   |     2     2      3      2      3 |
>   |----------------------------------|
>   |     3     1      3      2      1 |
>   +----------------------------------+
>
> .
> . exit
>
> end of do-file
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/