About Tom

Tom · ‎09-20-2011

Would need to know more about the coding scheme (or least a few more data points) but it looks like your number 110272 could be split into two pieces. 110 represents the year (add it to 1900) and 272 represents to number of days. 281 data _null_; 282 x=110272; 283 sas=mdy(1,1,1900+int(x/1000)) + mod(x,1000) -1 ; 284 diff = x - sas; 285 put x= sas= diff= sas date9.; 286 run; x=110272 sas=18534 diff=91738 29SEP2010

Tom · ‎09-20-2011

Why would you expect a dataset from the SASHELP library to necessarily exist in the PROJECT library? Change the query to match your situation. This is the clause that needs to change: where libname='SASHELP' and memname like 'CL%' You should change SASHELP to PROJECT to search for members in the PROJECT library. You should also change the rest of the conditional to find the particular datasets that you want. For example if you want to find every dataset in project then just leave it at: where libname='PROJECT'

Tom · ‎09-20-2011

Hopefully you have your answer with the INTCK function. In fact you should be able to get the number you want just from the SQL query. PROC SQL; SELECT inctck('QTR',MIN(enquiry_date),today()) INTO :N FROM sample; QUIT;

Tom · ‎09-20-2011

Try adding a WIDTH option to make the column wide enough for your longer title. You might also check what character you are using for you split character, but normally you would not use a space for that.

Tom · ‎09-20-2011

Depends on what you mean by "import". If you just have text files then I would just "input" them and would not bother with using any import functionality. 1) If the file names follow a simple pattern then just us a wildcard in the filename specification. data all ; infile '*.txt'; input column1 ; run; 2) If you have a static list of filenames then build a FILENAME statement. filename myfiles ('one.txt','two.txt,'three.txt','four.txt','five.txt'); data all; infile myfiles; input column1; run;

Tom · ‎09-19-2011

Why not generate the "title" lines in the report using ODS TEXT statements? I am not sure how that would interact with panels.

Tom · ‎09-19-2011

One way is to merge the two by MEMNO and compare the values. Because the variables have the same name in both dataset you will need to rename one set of names so that you can have two different variable names to compare. One way to avoid having to type all those new names is to create a dataset that looks like the history file, but has the address information from the master file. Then you could use PROC COMPARE to compare the two. data check ; merge demo demohistory(keep=memno) ; by memno; run; proc compare data=check compare=demohistory ; id memno; run;

Tom · ‎09-19-2011

You need to add a SEPARATED BY ' ' claus to the INTO cluase of the select statement. Otherwise SQL will only put the value from the first observation it pulls into the macro variable.

Tom · ‎09-19-2011

I think that you are asking if you can loop over the formatted values of variable in macro code. In general the answer is no, but there have been many discussions about ways to work around this. Here is one from earlier this year. http://listserv.uga.edu/cgi-bin/wa?A2=ind1103C&L=sas-l&D=0&P=24621

Tom · ‎09-19-2011

As data_null_ pointed out your ordering in your SQL select statement and your PROC REPORT are not the same. In SQL you order by: region,ordsale desc In your COLUMN statement you order by: region agent_info sale ordsale Try changing your COLUMN statement to : region ordsale agent_info sale

Tom · ‎09-18-2011

Here is a cleaned up version. I had skipped the line with the value date before. data company(keep=tran_no company_name db_cr_amount db curr appdt apptm tran_id rn payment_method trdt trtm valuedt fx_rate) company_account(keep=tran_no acctype acctid) ; infile src end=eof truncover; input @ ; if (_n_=1 or index(_infile_,'0c'x)) then do until (eof or _infile_=:'----'); input / @; end; do while (_infile_=:'----' or _infile_=' '); input / @ ; end; *-----------------------------------------------------------------------------; * Read transaction report ; *-----------------------------------------------------------------------------; input tran_no 1-8company_name $12-35 db_cr_amount : comma15. db $ curr $ appdt date7. apptm time8. / tran_id $ 1-9 rn $ 11 payment_method $ 13-64 trdt date7. trtm time8. / @12 valuedt date7. fx_rate & $20. @ ; format appdt trdt valuedt date9. apptm trtm time8. db_cr_amount comma15.2 ; output company; *-----------------------------------------------------------------------------; * Read the account numbers ; *-----------------------------------------------------------------------------; do until (_infile_=' ' or eof) ; input / @; if _infile_ ne ' ' then do; acctype = left(scan(_infile_,1,':')); acctid = left(scan(_infile_,2,':')); output company_account; end; end; run; proc print data=company width=min; run; proc print data=company_account width=min; run;

Tom · ‎09-17-2011

You could try this. It seems to read these two records. I tried to put in something to account for mulitple pages, but no way to test if it works. The account descriptions and ID numbers listed below each transaction where not consistently formatted between your two records, so I decided to output those into a separated dataset. I assumed that the TRAN_NO could serve as a key to link the records back together. data company(keep=tran_no company_name db_cr_amount db curr appdt apptm tran_id rn payment_method trdt trtm ) company_account(keep=tran_no acctype acctid) ; infile src end=eof truncover ; input @ ; if (_n_=1 or index(_infile_,'0c'x)) then do until(eof or _infile_=:'----'); input / @; end; if _infile_=:'----' or _infile_=' ' then input / @ ; *-----------------------------------------------------------------------------; * Read one company report ; *-----------------------------------------------------------------------------; input tran_no 1-8 company_name $12-35 db_cr_amount : comma15. db $ curr $ appdt date7. apptm time8. / tran_id $ 1-9 rn $ 11 payment_method $ 13-64 trdt date7. trtm time8. ; format appdt trdt date9. apptm trtm time8. db_cr_amount comma15.2 ; output company; *-----------------------------------------------------------------------------; * Read the account numbers ; *-----------------------------------------------------------------------------; do until (_infile_=' ' or eof) ; input / @; if _infile_ ne ' ' then do; acctype = left(scan(_infile_,1,':')); acctid = left(scan(_infile_,2,':')); output company_account; end; end; run; proc print data=company width=min; run; proc print data=company_account width=min; run;

Tom · ‎09-16-2011

# pointer will probably not work unless both your header lines and your report blocks have the same number of lines. If your blocks are consistently seven lines long then use / instead of # to move to the next line in your input. Then you can still scan the file line by line to look for the start of the next block. If they are not a consistent number of lines then it looks like you can read until you find the next empty line. You should look for start of pages. Usually you will see the formfeed character at the start of all pages (except the first). ( index(_infile_,'0C'x) ) You header line seems to be all dashes. To look for these I usually use this test : index(_infile_,'--') and compress(_infile_,'-')=' ' This will find lines that have at least two adjacent hyphens but no other printable characters. Another trick is let in the input statement parse the line into words for you and then based on what you see you can assign the right word your variables. infile ... truncover ; length x1-x20 $200 ; input x1-x20 ; if length(x1)=7 and . ^= input(x1,??6.) then *** found start of new block *** ;

Tom · ‎09-16-2011

try one of these three methods. data want; set have; by id date; if first.date; run; proc sql; create table want as select distinct id,date from have; quit; proc sort data=have out=want nodupkey ; by id date; run;

Tom · ‎09-16-2011

You should fix the data in the LIST dataset by doubling up the embedded quotes. So instead of 'bruegger's' you have it stored as 'bruegger''s'.

Online Status	Offline
Date Last Visited	yesterday