|
|  |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: Re: cumulative product, panel data
There are several ways you could do this. I think you discarded the
exp(sum(ln(x))) aproach too quickly. You could do this:
bys id (t): gen sumprod=exp(sum(ln(x)))
by id (t): gen sumzero=sum(x==0)
replace sumprod=0 if sumzero>0
drop sumzero
You could also have used the approach of the thread you found unhelpful by
realizing that, although you may not use -in 2/l- after a -by-, you can
use -if _n>1-
Michael Blasnik
----- Original Message -----
From: "Hamilton, David" <[email protected]>
To: <[email protected]>
Sent: Thursday, December 14, 2006 11:35 AM
Subject: st: cumulative product, panel data
I'm having difficultly with what seems a straightforward programming
exercise. I am trying to calculate the cumulative product for a variable
for panel data set.
My data basically consists of the following 3 variables: id, t, x.
id runs from 1 to 1,000; t (time interval) runs from 1 to 20; and x is a
ratio for each t (and id).
I am trying to calculate the cumulative product of x for each id across
time interval 1/20 in a new variable (prodx, say).
x does sometimes take the value zero (so exp(sum(ln(x))) isn't an
option).
I found the following thread unhelpful as "in" cannot be used with "by":
* * *
Assuming your original variable is x:
. generate runprod = x in 1
. replace runprod = runprod[_n-1]*x in 2/l
<snip>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
