About Tom

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I am not sure the &SYSJOBID will be unique over time so that you could use it as a historical reference.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;

Tom · ‎07-29-2011

I have started getting two copies of the email that is generated when some posts a reply to a discussion that I am following. Has anyone else noticed this?

Tom · ‎07-29-2011

It is not "stored" in an array. Arrays in SAS are just a tool to simplify referencing variables. You should use macro language constructs to build code that cannot be done with the normal language.

Tom · ‎07-29-2011

Does not look like a macro issue at all. You are just moving data around in the same row of a dataset. data want ; set have ; array v var1_1 - var1_12 ; diff = intck('month',date1,date2); do i = 12 by -1 while (i-diff > 0); v(i) = v(i-diff) ; end; do i = 1 to diff ; v(i) = 0; end; run; What is it that will vary in the macro call? The names of the array variables? The upper limit of 12?

Tom · ‎07-28-2011

You cannot ADD anything to dictionary.columns as it is just a view into the metadata about existing datasets. As I understand this you want to QUERY dictionary.columns and try to figure out which VARIABLES in your output dataset came from which input dataset. SAS just does basically a UNION of the variable names when it does a merge. So you just need to do a merge of the records from dictionary.columns. You can also create new columns when creating a new dataset so you should pull the information from that one also so that you can flag the variables that do not come from either of the two source datasets. Attributes like LENGTH will be set by the setting from the first dataset, but attributes like FORMAT have a more complex logic for how they are set. Plus your data step could have FORMAT or LABEL statements to change them. proc sql noprint ; create table ds1 as select * from dictionary.columns where libname = "IPSC" and memname = "AGENTPROFILE" order by name ; create table ds2 as select * from dictionary.columns where libname = "IPSC" and memname = "AGENTPROFILEADDITIONALINFO" order by name ; create table ds3 as select * from dictionary.columns where libname = "IPSC" and memname = "AGENTPROFILEWINFO" order by name ; quit; data varlist ; merge ds1 ds2 ; by name ; contribdataset = memname ; run; data varlist ; merge varlist ds3 ; by name ; if contribdataset = ' ' then contribdataset=memname; run; You could probably figure out how to do this in a single SQL statement.

Tom · ‎07-27-2011

Works in IE6 but not in Chrome. I am using Microsoft Windows XP Professional 5.1.2600 service pack 3.0. When I inspect the element in Chrome it shows this.

Tom · ‎07-27-2011

Nice idea, but the link doesn't seem to work. At least not in Chrome. When I click on that link it just inches up the screen a little bit rather than jumping to the top.

Online Status	Offline
Date Last Visited	Sunday