About Tom

Tom · ‎08-29-2011

It sounds like you want to split your ids into connected subgraphs. If you have SAS/OR licensed I have had success using PROC NETDRAW to do this. Otherwise you can code your own logic. I would start by converting your data from the current four variables to two variables and populate it with pairs of values from your four variables. data relations; set test; array x pan1-pan3 add1 ; do i=1 to dim(x)-1; do j=i+1 to dim(x); id1=x(i); id2=x(j); output; id1=x(j); id2=x(i); output; end; end; keep id1 id2 ; run; proc sort data=relations nodupkey; by id1 id2; run; You can grow your own subnets by looping and adding in all of the connected ids from the current list until the current subnet stops growing.

Tom · ‎08-24-2011

It does not look like the psftp program does what you want it to do. You can test that by taking SAS out of the problem and issuing the proposed command from a DOS prompt. I think it would be much easier to use SAS to write the script for psftp than to depend on an external file. But you could use an external file as a model for the script that SAS would write. The datastep below will replace all references to '%1' with the actual directory. %let sdir=d:\raj ; filename script temp; data _null_; infile 'abc.bat' ; file script; input; _infile_ = tranwrd(_infile_,'%1',"&sdir"); put _infile_; run; %sysexec psftp.exe username@hostname -p portnumber(22) -pw password -batch -b %sysfunc(pathname(script)) ;

Tom · ‎08-21-2011

Not finding the correct values is probably a result of the lines in your files being longer than the default line length for an INFILE statement. You should see notes in the log to the effect that some lines were truncated. Add a LRECL option to your INFILE statement such as LRECL=10000. Art's program is taking advantage of the fact that SAS can support a single wildcard (*) in a filename. You can test if it finds the files you want by issuing a DIR command with the same filename at the command line prompt on your PC. The INPUT statement without any variables will force SAS to read the line into the input buffer. You can reference this buffer using the special variable name _INFILE_. The SCAN function with a negative argument will scan backwards from the end of the string. You might want to specify what delimiters to use when scanning. For example change this to SCAN(_INFILE_,-1,' ') to force it to use space as the delimiter when scanning. Your example file was delimited with spaces and not tabs as you said above. The INPUT function will convert the characters pulled off the end of the line into a number using the BEST12. format. If you want to save the name of the individual file that a particular record came from in the output dataset then you will need to assign it to another variable as the variable named in the FILENAME= option of the INFILE statement will be automatically dropped from the output dataset.

Tom · ‎08-20-2011

Make sure that you have permission to run X command on the DIS server. Try running another program that does a different command. For example: x "echo hello > %sysfunc(pathname(work))\hello.dat" ; data _null_; infile "%sysfunc(pathname(work))\hello.dat"; input; list; run; If that works then try to find out if you have permission to run the cscript command on the server.

Tom · ‎08-20-2011

When this issue has been mentioned before it is usually related to the user not being allowed to send email from the process started by Windows Scheduler. You need to check how to enable access to your mailbox for sending email.

Tom · ‎08-12-2011

Not sure what you mean by truncation. I do not see any difference when running your code with or without the embedded titles.

Tom · ‎08-07-2011

Just add the TRUNCOVER option to your INFILE statement. See this recent thread. http://communities.sas.com/message/13199#13199

Tom · ‎08-02-2011

Many procs have NOPRINT options. You could also try re-directing listing to a null device. * Unix ; ods listing file='/dev/null' ; * PC ; ods listing file='nul:';

Tom · ‎08-02-2011

You need to standardize your variable names. The easiest is to use UPCASE or lowcase function. You also need to check if you have multiple records for the same variable. The extra data step below will do both. proc sort data=have1; by companyname tablename date year; run; data have2 ; set have1; by companyname tablename date year; if first.year then recno=0; recno + 1; var = lowcase(var); run; proc transpose data=have2 out=want (drop=_name_); by companyname tablename date year recno; var data; id var; run;

Tom · ‎08-02-2011

Looks like you have two values of VAR that SAS wants to convert to the same variable name. Such as 'CASH' and 'cash', 'Net_earnings' and 'Net_Earnings'. Do a proc freq on the variable VAR and see what you have.

Tom · ‎08-02-2011

Are you trying to point the libref to a different location? Or are you just re-running the code that assigned the libref? If the latter then test first before defining the library. Here is can example command using the IFC and LIBREF functions to conditionally issue the libname statement only when the libref does not already exist. 4 %sysfunc(ifc(%sysfunc(libref(macros)),libname macros '~/temp',)); NOTE: Libref MACROS was successfully assigned as follows: Engine: V9 Physical Name: /home/abernt/temp 5 %sysfunc(ifc(%sysfunc(libref(macros)),libname macros '~/temp',)); The other thing to consider is changing the SASMSTORE option before trying to modify the libref.

Tom · ‎08-02-2011

If you are embedding the titles why not just start your titles with title3 ? title1 ' '; title3 "Summary of Date"; title4 "as of ";

Tom · ‎08-02-2011

What evidence do you have that it did not sort properly? Can you give more indication of what you need to do with the data? You requirements seem to be in conflict. If you really need to access the file in both DATE/PATIENT and PATIENT/DATE order then you could make two copies of the data. Or use one or more indexes. With indexes you can access the data in the index order without physically sorting the records.

Tom · ‎08-02-2011

I am not sure the &SYSJOBID will be unique over time so that you could use it as a historical reference. I know that on Unix it will restart the process id numbers when the server reboots. They might even cycle around if you run the server long enough. Also what if your jobs are distributed across servers.

Tom · ‎07-30-2011

In general you want to sort and keep the last one per group. proc sort data=have out=want ; by code date volume; run; data want; set want; by code date ; if last.date; run;

Online Status	Offline
Date Last Visited	Monday