I assume you have numeric date nd, and compute mean cnv across
individuals as, e.g.
. egen mcnv=mean(cnv), by nd
If you have data on every day, or you don't care if any are missing,
and you really want the prior 7 calendar days, you could start with
. bys nd: gen byte nfirst=(_n!=1)
. sort nfirst nd
. set obs `=_N+7'
. replace nd=0 if nd==.
. g byte s=0
. forval d=1/7 {
replace scnv=max(0,scnv)+ /*
*/ max(0,mcnv[_n-`d']*inrange(nd[_n-`d'],nd-7,nd-1)) if nfirst==0
}
. g byte ccnv=0
. forval d=1/7 {
replace ccnv=ccnv+(mcnv[_n-`d']!=. & inrange(nd[_n-`d'],nd-7,nd-1)!=0)
}
. g m7cnv=scnv/ccnv
. li, noo clean
d nd mcnv scnv ccnv m7cnv
. 0 . 0 0 .
. 0 . 0 0 .
. 0 . 0 0 .
. 0 . 0 0 .
. 0 . 0 0 .
. 0 . 0 0 .
. 0 . 0 0 .
26nov2003 16035 6 0 0 .
24aug2004 16307 32 0 0 .
26aug2004 16309 27 32 1 32
27aug2004 16310 30 59 2 29.5
28aug2004 16311 34 89 3 29.66667
29aug2004 16312 5 123 4 30.75
30aug2004 16313 30 128 5 25.6
31aug2004 16314 3 158 6 26.33333
01sep2004 16315 27 129 6 21.5
02sep2004 16316 43 156 7 22.28572
03sep2004 16317 12 172 7 24.57143
04sep2004 16318 2 154 7 22
05sep2004 16319 21 122 7 17.42857
06sep2004 16320 44 138 7 19.71428
07sep2004 16321 26 152 7 21.71428
Though there may be a one-line solution avail after much thought...
On 8/17/05, James Avery <[email protected]> wrote:
> Dear all,
> I would like to create a new variable, prior
> week campaign news volume (PWCNV) that is the mean CNV
> for the prior 7 days. Is it possible to do this with
> the egen mean function or with another egen function?
*
* 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/
