<>
" Incidentally, would using "double" have worked for a larger number of
digits, or did this function give me just enough extra precision to
accomplish the current task?"
I would say your data setup provides a laboratory to find out:
*******
clear*
inp byte(itemid cityid1 cityid2)
22 61 6
22 61 8
22 61 9
22 61 7
22 61 10
22 61 13
end
gen long newid1 = 1000000*itemid + 1000*cityid1 + cityid2
gen long newid2 = 10000000*itemid + 1000*cityid1 + cityid2
gen double newid3 = 100000000*itemid + 1000*cityid1 + cityid2
gen double newid4 = 100000000000*itemid + 1000*cityid1 + cityid2
format newid? %20.0f
list newid?, noobs
*******
HTH
Martin
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Christopher
Hajzler
Sent: Sonntag, 6. Dezember 2009 21:38
To: [email protected]
Subject: Re: st: RE: problem creating unique identifier
Thank you, Martin. I was could not imagine when I initially
discovered the problem that there was a standard reason for what was
going on, but I get it now (on a basic level). Using "double" seems
to be working out fine, and I really appreciate your help.
Incidentally, would using "double" have worked for a larger number of
digits, or did this function give me just enough extra precision to
accomplish the current task?
Cheers,
Chris
On Sun, Dec 6, 2009 at 1:59 PM, Martin Weiss <[email protected]> wrote:
>
> <>
>
> See also http://www.stata.com/support/faqs/data/prec.html, NJC`s
> http://www.stata-journal.com/sjpdf.html?articlenum=dm0022 and the entry
> "Precision of numeric storage types" in -help data_types-...
>
>
> HTH
> Martin
>
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of Christopher
> Hajzler
> Sent: Sonntag, 6. Dezember 2009 20:47
> To: [email protected]
> Subject: st: problem creating unique identifier
>
> Dear Statalist,
>
> My version of Stata is doing something quite strange (Version 8.0
> intercooled). I have several variables which are contained in
> separate data files, and I want to merge them by creating a unique
> identifier for each observation (over which I can sort the data - I
> realize it is not strictly necessary to do it this way, but I thought
> this method would help avoid errors). Each observation represents a
> city pair (just over 100 cities), and an "item", and in each file I've
> created the variable using the following command:
> gen newid = 0
> replace newid = 1000000*itemid + 1000*cityid1 + cityid2
>
> For some reason Stata will only do a partial job - it seems to be
> getting the itemid and cityid1 right when producing newid (the first 4
> or 5 digits), and "almost" gets cityid2 right, assigning the correct
> values every few observations but then incorrectly allocating the same
> value to adjacent observations. For example, one file reads:
>
> newid itemid cityid1 cityid2 var1
> ...
> 22061006 22 61 6 ...
> 22061008 22 61 8 ...
> 22061008 22 61 9 ...
> 22061008 22 61 7 ...
> 22061010 22 61 10 ...
> 22061012 22 61 13 ...
> ...
>
> I have worked with much larger datasets, so I cannot imagine memory
> allocation is an issue. Any other ideas?
>
> Best wishes,
> Chris
>
> *
> * 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/
>
>
> *
> * 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/
>
*
* 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/
*
* 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/
