Kevin McConeghy <kevinmcconeghy@gmail.com> asked,
> I have a dataset with ~2.2 mill obs like so:
>
> id stringvar + other variables
> 1 x
> 1 y
> 1 z
> 2 a
> 3 d
> 4 g
> 4 h
>
> [...]
> I was trying to combine the stringvar to collapse and make id a unique
> key, like so:
>
> id stringvar
> 1 xyz
> 2 a
> 3 d
> 4 gh
>
> [...] [-reshape- ran out of memory] [...]
>
>
> Is there some way to skip the reshape step [...]?
Here is my solution. First, let me set up the toy problem,
. clear all
. input id str1 stringvar
id stringvar
1. 1 x
2. 1 y
3. 1 z
4. 2 a
5. 3 d
6. 4 g
7. 4 h
8. end
My solution is,
. sort id
. gen str result = ""
. by id: replace result = result[_n-1] + stringvar
. by id: keep if _n==_N
Below I run that, with a few -list-s added:
. sort id
. gen str result = ""
(7 missing values generated)
. by id: replace result = result[_n-1] + stringvar
(7 real changes made)
. list
+------------------------+
| id string~r result |
|------------------------|
1. | 1 x x |
2. | 1 y xy |
3. | 1 z xyz |
4. | 2 a a |
5. | 3 d d |
|------------------------|
6. | 4 g g |
7. | 4 h gh |
+------------------------+
. by id: keep if _n==_N
(3 observations deleted)
. list
+------------------------+
| id string~r result |
|------------------------|
1. | 1 z xyz |
2. | 2 a a |
3. | 3 d d |
4. | 4 h gh |
+------------------------+
In my solution,
. sort id
. gen str result = ""
. by id: replace result = result[_n-1] + stringvar
. by id: keep if _n==_N
watch out for the first line, -sort id-. It should really read,
. sort id some_other_variable
We need to specify the order within equal values of id to make
the the order of the letters deterministic. Perhaps Kevin want
the letters is alphabetical order, in which case -sort id- should
change to -sort id stringvar-.
-- Bill
wgould@stata.com
*
* 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/
