Hi.Patrick is right. To avoid to overwrite F8,you need to change the code a little bit to retain the changed value of F8. [pre] %let mvar1= rahim1 ramu1 sham1; %let mvar= rahim ramu sham; %let val=3; data testing22; rahim="good"; ramu="cool"; sham="fine"; rahim1="rahim"; ramu1="ramu"; sham1="akram"; a="from rahim";output; a="from ramu";output; a="from akram";output; a="from aram";output; a="from &sham";output; a="from &sham;";output; rahim="good";output; ramu="cool";output; sham="fine";output; rahim1="rahim";output; ramu1="ramu";output; sham1="akram";output; run; options mprint; %MACRO Q4; data testing222; set testing22; F8=a; %do i=1 %to &val; %LET m=%SCAN("&mvar",&i," ") ; %LET k=%SCAN("&mvar1",&i," ") ; F8=TRANWRD(F8,&k,&m); %end; run; %MEND; %q4 [/pre] Ksharp ... View more

Haha. I think I found the answer.See the data below. [pre] 17 20041016 20041107 1010 N 17 20041016 20050131 1010 Y 17 20041016 20050131 1010 N 17 20050718 20050919 1010 N 17 20050718 20060221 1010 Y 17 20070206 20070314 1227 N 17 20070206 20071021 1227 N 17 20080714 20081027 1624 Y [/pre] For third record,which has the same id prod with the last observation,but they have different value of outc, one is Y ,another is N. So run the Art's code ,it also get three obs,But my code is four obs. So we need to promote the proc sort. [pre] proc sort data=idprod; by id prod date2 outc; run; [/pre] Ksharp ... View more

>Obviously it works to place PROC SQL inside a data step, but to me it seems a little odd. Hi. That is not insert proc sql, I forget RUN; and thank SAS to intelligently identify data-step and proc sql. 'b.*' stands for select any variable from idprod,not for the new alias. I am not sure why there is different result for Me and Art.acctually i do not understand art's code totally. the following is the code i change a little bit. Indeed,You can use 'merge idprod index; by...' and 'where .....' to instead of proc sql. [pre] data idprod; input id DATE1 :YYMMdd10. DATE2 :YYMMdd10. PROD OUTC $; format date1 date2 yymmddn8.; datalines; 17 20041016 20041107 1010 N 17 20041016 20050131 1010 Y 17 20050718 20050919 1010 N 17 20050718 20060221 1010 Y 17 20070206 20070314 1227 N 17 20070206 20071021 1227 N 17 20080714 20081027 1624 Y RUN; proc sort data=idprod; by id prod date2; run; data index(where=(count eq 1 and outc eq 'Y')); set idprod; by id prod; if first.prod then count=0; if outc='Y' then count+1; * to identify the first occurrence of 'Y'; run; proc sql feedback; create table selected_records as select b.id as _id,b.prod as _prod,b.date1 as _date1,b.date2 as _date2,b.outc as _outc from index as a left join idprod as b on a.id = b.id and a.prod = b.prod where b.date2 le a.date2; quit; proc print noobs;run; [/pre] Ksharp ... View more

Hi. I am not sure.But the code is worked fine for me. [pre] data _null_; date=input('12aug2009:10:20:32',anydtdtm22.); format date datetime9.; put date=; run; [/pre] Ksharp ... View more

Ok. I finally find it. But forget also need 'proc gprint'; the following is the code i steal from somewhere. [pre] filename logtxt 'c:\temp\report.txt'; goptions reset=all; proc printto print=logtxt; run; proc report data=sashelp.class nowd; run; proc printto; run; proc gprint fileref=logtxt name='logGraph'; run; [/pre] Ksharp ... View more

Hi. I think your logic is too complex.and data-step conjoint with proc sql is perfect to process data. [pre] data idprod; input id DATE1 :YYMMdd10. DATE2 :YYMMdd10. PROD OUTC $; format date1 date2 yymmddn8.; datalines; 17 20041016 20041107 1010 N 17 20041016 20050131 1010 Y 17 20050718 20050919 1010 N 17 20050718 20060221 1010 Y 17 20070206 20070314 1227 N 17 20070206 20071021 1227 N 17 20080714 20081027 1624 Y RUN; proc sort data=idprod; by id prod date2; run; data index(where=(count eq 1 and outc eq 'Y')); set idprod; by id prod; if first.prod then count=0; if outc='Y' then count+1; * to identify the first occurrence of 'Y'; proc sql feedback; create table selected_records as select b.* from index as a left join idprod as b on a.id = b.id and a.prod = b.prod and b.date2 le a.date2; quit; proc print noobs;run; [/pre] Ksharp ... View more

Hi.Cynthia. That maybe you have null label for variable name which SAS treat is as the missing value of numeric , so assign a label 'name', That Warning would not appear again. Or you can color it with the same background's color to hidden. [pre] proc format; value $act 'Actual' = 'purple' 'Forecast' = 'white' other = 'pink'; value $a 'na'='pink' other='cx888888'; run; ods listing close; ods html file='span_example2.html' style=sasweb; proc report data=sashelp.class(obs=5) nowd style(header)={background=$act. foreground=$a.}; column ('na ' name) ('Actual' sex age) ('Forecast' height weight); define name / display 'name '; define sex / display; define age / display; define height / mean; define weight / mean; run; ods html close; [/pre] Ksharp Message was edited by: Ksharp ... View more

Yes. You can refer to ' proc printto ' which can change the result of proc report into jpg that looks like exactly the output window of SAS. Ksharp ... View more

Hi.I found the code I writed before. But not test it, You can refer to it. [pre] %let subdir=D:\sasdata\; filename dir pipe "dir &subdir.*.xls /B"; data new; infile dir truncover end=last; input filename $100.; filename=cats("&subdir",filename); call symputx(cats('path',_n_),filename); call symputx(cats('dsn',_n_),scan(scan(filename,3,'\'),2,'_')); if last then call symputx('nobs',_n_); run; %put _user_; %macro import; %do i=1 %to &nobs; proc import datafile="&&path&i" out=&&dsn&i dbms=excel replace; getnames=yes; run; %end; %mend import; %import [/pre] Ksharp ... View more

Or . You need one more statement to release your libname. [pre] libname macros clear; [/pre] Ksharp ... View more

Hehe.Ok you can mail to me. And also I think @data _null_; can do it better than me.That is just to post my solution before @data _null_; saw it. Ksharp ... View more

Ok. I can not understand enough for your said. Can you point out the proper formate to read this style of datetime , I do not find the informat after look up documentation for a long time. Ksharp ... View more

Hi. I also forget the order of date is very important. Emmmm.I think that would be more simple. 🙂 [pre] data temp; input id invitedate ddmmyy10.; cards; 1 01/01/2010 1 12/01/2010 1 15/05/2010 2 01/01/2010 2 15/05/2010 2 19/06/2010 2 30/07/2010 ; run; proc freq data=temp; tables id*invitedate /list out=result(drop=percent); format invitedate monyy5.; run; proc print noobs label; label count='invites'; run; [/pre] Ksharp ... View more

Although you do not want proc transpose (the best choice), then @data step will meet your taste. [pre] data b; input custid emp $ age; cards; 01 emp1 20 01 emp2 30 01 emp3 40 02 emp1 50 02 emp2 60 02 emp3 70 03 emp1 80 03 emp2 90 03 emp3 20 ; run; proc sort data=b; by custid; run; data op(drop=count emp age); set b; by custid; array e{3} emp1 emp2 emp3; retain emp1 emp2 emp3; if first.custid then count=0; count+1; e{count} = age; if last.custid then output; run; proc print noobs;run; [/pre] Ksharp ... View more

OK. That would not be too complicated. [pre] filename csv 'c:\test.csv'; data temp; ; infile csv delimiter = ',' lrecl=32767 firstobs=1; informat datetime $30.; input datetime; _datetime=input(catx('/',put( input(scan(datetime,1,' '),ddmmyy10.) ,date9.),put( input( catx(' ',scan(datetime,2,' '),scan(datetime,3,' ')) ,anydttme11. ) ,timeampm11.)),datetime22.); format _datetime dateampm22. ; keep _datetime; run; proc print noobs;run; [/pre] Ksharp ... View more