st: average value among differing numbers of variables

[Radu Ban <rban@nber.org>]

Dear listers,

This is a data management question. The data that I'm looking at (daily 
U.S. weather) has the following structure.

	day1	flag1	day2	flag2	day3	flag3	day4	flag4	day5	flag5	...	day31	flag31
	0	s	a2	a	0	s	0	s	a5	a		a31		
	0	s	0	s	b3	a	b4		b5			b31
	c1		0	s	0	s	0	s	c5	a		c31

the "s" flag means that the measured element (say inches of rain) is 
accumulated over those days, which are assigned a 0 value, and the 
accumulated amount is reported in the day flagged with "a". i would like to 
replace the 0 value for the accumulation days with the average of the 
accumulated value over those days.

given the notations above, specifically, i would like to replace 0, 0, b3 
(in the second row) with b3/3; 0, 0, 0, c5 (in the third row) with c5/4, 
and so on. note that, as in the first row there can be more than one 
accumulation series per row.

i figured out that each type of accumulation, a_ij(starting at day i ending 
at day j) must be identified, so that in the end i can use:

forval j = 2/31 {
	forval i = 1/`j' {
		egen daymean = rmean(day`i'-day`j') if a_`i'`j' == 1
		replace day`i' = daymean
		drop daymean
	}
}

but i'm not sure how to define all a_ij

i hope my question is clear. any help is greatly appreciated.

thank you,

Radu Ban


*
*   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/