st: RE: AW: RE: AW: AW: Creating a Group Pair ID (where the generating variables order shouldn't matter)

["Nick Cox" <n.j.cox@durham.ac.uk>]

Good question. 

Your identifier would lead to an integer variable with labels if -group()- were used with the -label- option. A good identifier should be informative as well as distinct, so I regard using -label- as very good practice. I didn't spot that you weren't following that very good practice. My mistake. 

Nick 
n.j.cox@durham.ac.uk 

Martin Weiss

" Which could in turn be made simpler:"


Though the two approaches hardly lead to the same result. My notion of an
"ID", as originally requested, would not be a string such as "England
France", but a numeric variable running from 1 to N, with N the number of
distinct groups.

*************
clear*
inp str20 c1id str20 c2id
"US" "Canada"
"US" "Mexico"
"Canada" "US"
"US" "France"
"France" "England"
"France" "US"
end

gen newid = cond(c1id < c2id, c1id, c2id)  /* 
*/ + " " + cond(c2id < c1id, c1id, c2id)

sort newid

l, sepby(newid) noo
*************


What makes you think that my approach returns a " ...integer variable with
labels."? All I can find is a -varlabel- attached to my newid.




. d newid

              storage  display     value
variable name   type   format      label      variable label
----------------------------------------------------------------------------
---------------
newid           float  %9.0g                  group(first second)




HTH
Martin


-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Nick Cox
Gesendet: Freitag, 23. Juli 2010 13:59
An: statalist@hsphsun2.harvard.edu
Betreff: st: RE: AW: AW: Creating a Group Pair ID (where the generating
variables order shouldn't matter)

Which could in turn be made simpler: 

gen first = cond(c1id < c2id, c1id, c2id)
gen second = cond(c2id < c1id, c1id, c2id)
egen newid = group(first second)
drop first second
sort newid

could become 

gen newid = cond(c1id < c2id, c1id, c2id) + " " + cond(c2id < c1id,
c1id, c2id)
sort newid

The cost is greater storage, which may or may not bite: that is, -newid-
is a string variable rather than an integer variable with labels. But if
you have enough space to create -first- and -second- as string
variables, even fleetingly, you presumably have enough space for a
string -newid-. 

Nick 
n.j.cox@durham.ac.uk 

Martin Weiss

Essentially, the technique advocated in NJC`s tip boils down to a simple
trick:

*************
clear*
inp str20 c1id str20 c2id
"US" "Canada"
"US" "Mexico"
"Canada" "US"
"US" "France"
"France" "England"
"France" "US"
end

gen first = cond(c1id < c2id, c1id, c2id)
gen second = cond(c2id < c1id, c1id, c2id)

egen newid = group(first second)

drop first second
sort newid

l, sepby(newid) noo
*************

Martin Weiss

Try NJC`s http://www.stata-journal.com/article.html?article=dm0043

J Taylor

I am trying to create an ID corresponding to numbers from two lists.
For example, if the two lists were of countries, one would have

clear
input str20 c1id str20 c2id
"US" "Canada"
"US" "Mexico"
"Canada" "US"
"US" "France"
"France" "England"
"France" "US"
end
egen newid = group(c1id c2id)

I would like newid to create an ID pair for each country pair.  My
first instinct was to use the egen group command.  However, the
problem is that egen group takes into account which id comes first.
For example, (c1id,c2id)=(United States,Canada) and (c1id,c2id)=(
Canada ,United States) have different IDs.  I would like them to be
able to have the same ID.  That is, I would like to create newid as a
group pair ID, reflecting which two countries are in the pair, and
where the order doesn't matter.


*
*   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/