如何在STATA中解决这样的难题

公司代码	分析师预测时间	预测终止时间
1	2005-5-9	2005-12-31
1	2005-5-9	2006-12-31
1	2005-5-9	2007-12-31
1	2005-7-31	2005-12-31
1	2005-7-31	2006-12-31
1	2005-7-31	2007-12-31
1	2005-8-19	2005-12-31
1	2005-8-19	2006-12-31
1	2005-8-19	2007-12-31
1	2005-9-29	2005-12-31
1	2005-9-29	2006-12-31
1	2005-11-1	2005-12-31
2	2005-11-1	2006-12-31
2	2005-11-1	2007-12-31
2	2005-11-6	2006-12-31
2	2005-11-6	2007-12-31
2	2005-11-6	2008-12-31
2	2006-1-11	2006-12-31
2	2006-2-22	2006-12-31
2	2006-3-8	2006-12-31
2	2006-4-4	2006-12-31
3	2006-4-4	2007-12-31
3	2006-4-4	2008-12-31
3	2006-4-10	2006-12-31
3	2006-4-10	2007-12-31
3	2006-4-19	2006-12-31
3	2006-4-19	2007-12-31
3	2006-4-27	2006-12-31
4	2006-4-27	2007-12-31
4	2006-4-27	2008-12-31
4	2006-5-28	2006-12-31
4	2006-5-28	2007-12-31
4	2006-5-30	2006-12-31
4	2006-5-30	2007-12-31
4	2006-6-12	2006-12-31
4	2006-6-12	2007-12-31
4	2006-6-12	2008-12-31
4	2006-6-15	2006-12-31
4	2006-6-15	2007-12-31

我现在想解决的问题是我想知道每一个公司当年有多少分析师跟踪，比如说发布时间是2005年，终止时间为2005年，则这个分析师对该公司跟踪时间为2005年，如果发布时间为
2005年，终止时间为2006年，则这个分析师对公司的跟踪为2005和2006年，现在有很多分析师发布不同的跟踪和终止时间，如何求出每一个公司每年的的跟踪人数？

试试：
clear
input stkcd str10 dbgn str10 dend
1 2005-5-9 2005-12-31
1 2005-5-9 2006-12-31
1 2005-5-9 2007-12-31
1 2005-7-31 2005-12-31
1 2005-7-31 2006-12-31
1 2005-7-31 2007-12-31
1 2005-8-19 2005-12-31
1 2005-8-19 2006-12-31
1 2005-8-19 2007-12-31
1 2005-9-29 2005-12-31
1 2005-9-29 2006-12-31
1 2005-11-1 2005-12-31
2 2005-11-1 2006-12-31
2 2005-11-1 2007-12-31
2 2005-11-6 2006-12-31
2 2005-11-6 2007-12-31
2 2005-11-6 2008-12-31
2 2006-1-11 2006-12-31
2 2006-2-22 2006-12-31
2 2006-3-8 2006-12-31
2 2006-4-4 2006-12-31
3 2006-4-4 2007-12-31
3 2006-4-4 2008-12-31
3 2006-4-10 2006-12-31
3 2006-4-10 2007-12-31
3 2006-4-19 2006-12-31
3 2006-4-19 2007-12-31
3 2006-4-27 2006-12-31
4 2006-4-27 2007-12-31
4 2006-4-27 2008-12-31
4 2006-5-28 2006-12-31
4 2006-5-28 2007-12-31
4 2006-5-30 2006-12-31
4 2006-5-30 2007-12-31
4 2006-6-12 2006-12-31
4 2006-6-12 2007-12-31
4 2006-6-12 2008-12-31
4 2006-6-15 2006-12-31
4 2006-6-15 2007-12-31
end

gen db = date(dbgn, "YMD")
format db %td
gen de = date(dend, "YMD")
format de %td
gen yb = year(db)
gen ye=year(de)

sort stkcd
forval i = 2005/2009 {
        by stkcd: egen n_`i' = total(cond(`i'>=yb & `i' <=ye), 1, 0)
}

学习了。谢谢上面的楼主

*设begin与end都是字符型变量
forv i=2005/2009{
g  y`i'=(real(substr(begin,1,4))<=`i')&(real(substr(end,1,4))>=`i')
}
collapse (sum) y2005-y2009,by(id)

sungmoo 发表于 2010-1-11 23:22
*设begin与end都是字符型变量
forv i=2005/2009{
g  y`i'=(real(substr(begin,1,4))=`i')
}
collapse (sum) y2005-y2009,by(id)

用collapse很巧妙啊，学习啦。