This seams to work. Notice that lag2/lag3 are pseudo lags by virtue of calculation after the output statement. You will probably want to drop unwanted variables.

Regards,

Gregg (wishing I were at SUGI) Snell

data test; input ID YEAR X1 X2; datalines; 1 1990 0.3 10 1 1991 0.45 15 1 1992 0.4 14.5 1 1993 0.25 15 1 1994 0.5 20 3 1999 1.1 50 3 2000 1.3 100 3 2001 2.45 150 3 2002 3.4 145 3 2003 2.5 75 3 2004 4.5 90 ; run;

data test2; set test; by id; lag1x1=lag1(x1); retain lag2x1 lag3x1; if first.id then do; lag1x1=.; lag2x1=.; lag3x1=.; end; lead1=_n_+1; lead2=_n_+2; lead3=_n_+3; if lead1 le maxobs then set test(keep=id x1 rename=(id=lead1id x1=lead1x1)) nobs=maxobs point=lead1; else lead1x1=.; if lead1id ne id then lead1x1=.; if lead2 le maxobs then set test(keep=id x1 rename=(id=lead2id x1=lead2x1)) nobs=maxobs point=lead2; else lead2x1=.; if lead2id ne id then lead2x1=.; if lead3 le maxobs then set test(keep=id x1 rename=(id=lead3id x1=lead3x1)) nobs=maxobs point=lead3; else lead3x1=.; if lead3id ne id then lead3x1=.; output; lag3x1=lag2x1; lag2x1=lag1x1; run;

Ed Heaton wrote:

>Chang, > >I guess I didn't read his email closely enough. > >No, it doesn't work for lags within BY groups. Sorry for not following >the specs. > >Ed > >Edward Heaton, SAS Senior Systems Analyst, >Westat (An Employee-Owned Research Corporation), >1600 Research Boulevard, RW-3541, Rockville, MD 20850-3195 >Voice: (301) 610-4818 Fax: (301) 610-5128 >mailto:EdHeaton@Westat.com http://www.Westat.com > > > >-----Original Message----- >From: Chang Chung [mailto:chang_y_chung@HOTMAIL.COM] >Sent: Tuesday, April 12, 2005 8:46 AM >To: Ed Heaton >Subject: Re: lag & lead > > >On Tue, 12 Apr 2005 08:12:46 -0400, Ed Heaton <EdHeaton@WESTAT.COM> >wrote: > >Hi, Ed, > >Does this work for lagging or leading within a by-group? -- which Thomas >needs? > >Cheers, >Chang > > > >>Data new ; >> Set test ; >> If ( _n_ gt 1 ) then set test( >> keep=X1 >> rename=( X1=LAG1X1 ) >> ) ; >> If ( _n_ gt 2 ) then set test( >> keep=X1 >> rename=( X1=LAG2X1 ) >> ) ; >> If ( _n_ gt 3 ) then set test( >> keep=X1 >> rename=( X1=LAG3X1 ) >> ) ; Merge >> test( >> firstObs=2 >> keep=X1 >> rename=( X1=LEAD1X1 ) >> ) >> test( >> firstObs=3 >> keep=X1 >> rename=( X1=LEAD2X1 ) >> ) >> test( >> firstObs=4 >> keep=X1 >> rename=( X1=LEAD3X1 ) >> ) >> ; >>Run ; >> >>This can be made shorter with a couple of macros. >> >>%macro lookAhead( var= , data=_last_ , phase=1 ) ; >>/* This macro must be called inside a MERGE statement. */ >> &data( >> firstObs=%eval( &phase + 1 ) >> keep=&var >> rename=( &var=LEAD&phase&var ) >> ) >>%mEnd lookAhead ; >> >>%macro lookBehind( var= , data=_last_ , phase=1 ) ; >> If ( _n_ gt &phase ) then set &data( >> keep=&var >> rename=( &var=LAG&phase&var ) >> ) ; >>%mEnd lookBehind ; >> >>Options mergeNoBy=noWarn ; >>Data new ; >> Set test ; >> %lookBehind( var=X1 , data=test , phase=1 ) >> %lookBehind( var=X1 , data=test , phase=2 ) >> %lookBehind( var=X1 , data=test , phase=3 ) >> Merge >> %lookAhead( var=X1 , data=test , phase=1 ) >> %lookAhead( var=X1 , data=test , phase=2 ) >> %lookAhead( var=X1 , data=test , phase=3 ) >> ; >>Run ; >>Options mergeNoBy=noWarn ; >> >>Ed >> >>Edward Heaton, SAS Senior Systems Analyst, >>Westat (An Employee-Owned Research Corporation), >>1600 Research Boulevard, RW-3541, Rockville, MD 20850-3195 >>Voice: (301) 610-4818 Fax: (301) 610-5128 >>mailto:EdHeaton@Westat.com http://www.Westat.com >> >> >> >>-----Original Message----- >>From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of >>Thomas >>Sent: Tuesday, April 12, 2005 4:04 AM >>To: SAS-L@LISTSERV.UGA.EDU >>Subject: lag & lead >> >> >>Hi all, >> >>I know the "lag" function is available to lag the variable for >>N-period. But, I have a problem on how to take the "forward/leading" of >> >> > > > >>the variable using SAS. I hereby illustrate my problem with the >>follwoing example. I would very much appreciate it if you could kindly >>advise me how to achieve it (A generic case). >> >>Thank you very much! >>Thomas >> >>*Input; >>ID YEAR X1 X2 >>1 1990 0.3 10 >>1 1991 0.45 15 >>1 1992 0.4 14.5 >>1 1993 0.25 15 >>1 1994 0.5 20 >>3 1999 1.1 50 >>3 2000 1.3 100 >>3 2001 2.45 150 >>3 2002 3.4 145 >>3 2003 2.5 75 >>3 2004 4.5 90 >> >>*Expected output (lag & forward of X1); >>ID YEAR X1 X2 LAG1X1 LAG2X1 LAG3X1 LEAD1X1 LEAD2X1 >>LEAD3X1 >>1 1990 0.3 10 . . . 0.45 0.4 0.25 >>1 1991 0.45 15 0.3 . . 0.4 0.25 0.5 >>1 1992 0.4 14.5 0.45 0.3 . 0.25 0.5 . >>1 1993 0.25 15 0.4 0.45 0.3 0.5 . . >>1 1994 0.5 20 0.25 0.4 0.45 . . . >>3 1999 1.1 50 . . . 1.3 2.45 3.4 >>3 2000 1.3 100 1.1 . . 2.45 3.4 2.5 >>3 2001 2.45 150 1.3 1.1 . 3.4 2.5 4.5 >>3 2002 3.4 145 2.45 1.3 1.1 2.5 4.5 . >>3 2003 2.5 75 3.4 2.45 1.3 4.5 . . >>3 2004 4.5 90 2.5 3.4 2.45 . . . >> >>/*The output only show for the "lag and lead" of one variable. may I >>know how to generalize to N variable case and the N "lag and lead"?*/ >> >> > > > >