Hi Jason,
How about
drop if quantity==0
bysort sid pid bid (year): keep if _n==1
for (1), and,
bysort sid pid year (quantity) : keep if _n==1
gen answer=year if quantity>=50
bysort sid pid (answer): keep if _n==1
for (2)?
For (1) you can change the order of the two commands. But I suppose
restricting sample first is faster (but insignificantly maybe).
Ichiro
On 05.2.24 13:46, "Jason Hwang" <[email protected]> wrote:
> Dear Statalist Users,
>
> I have panel data and I would like to extract certain information for a
> survival analysis. Conceptually, I have data on supplier, buyer and
> quantity sold and I want to ask: when is the first date at which buyer
> buys more than x amount.
>
> My data is of the following form:
>
> year sid pid bid quantity
> 1990 111 1 555 0
> 1990 111 1 777 10
> 1990 111 2 555 100
> 1990 111 2 777 0
> 1991 111 1 555 3
> 1991 111 1 777 25
> 1991 111 2 555 5
> 1991 111 2 777 5
>
> sid: seller id
> pid: product id
> bid: buyer id
>
> So this tells me for each seller, how much product they sold to which
> buyer. Now I want to ask two questions:
>
> (1) For each seller and for each product and buyer, when is the first year
> at which quantity sold is positive?
>
> So the output I would want would be:
>
> sid pid bid answer
> 111 1 555 1991
> 111 1 777 1990
> 111 2 555 1990
> 111 2 777 1991
>
> The complication is that in the real data, quantity sold may become
> positive, zero, then again positive. How can I only pick up the first time
> it becomes positive?
>
> With this simple data, it's relative easy. I can just sort by sid pid bid
> and pick the first year that comes up but I was wondering if there's a
> more elegant way that's also quicker.
>
> (2) This is trickier. For each selle and product, I want the first year at
> which quantity sold is at least x for all buyers. For example, say x=50.
> mThen my answer would be
>
> sid pid answer
> 111 1 .
> 111 2 .
>
> since in neither year does the seller sell more than 50 to ALL buyers.
>
> However if x=5, then the output woud be
>
> sid pid answer
> 111 1 .
> 111 2 1991
>
> I would appreciate any hints. Any references or tricks for getting this
> type of panel data ready for survival-type analysis would be helpful as
> well. Thank you very much for your help.
>
> Jason Hwang
> *
> * 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/
>
*
* 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/
