Running SAS on Unix in Batch mode...

Alternative to Interactive SAS discussed @ http://sastechies.blogspot.com/2010/02/running-sas-on-unix-interactively-using.html one can use SAS in batch mode.

SAS in batch mode…

Create a .sas file and pasted the above code at your desired location on your unix box….if you are using PC then create the file and ftp it to your home drive (i.e. ~/)

The batch SAS command would be –

/opt/sas/sasb /pathtosasprog/prog.sas –log /pathtologfile/prog.log –print /pathtooutput/prog.lst

I have put the prog.sas in my home directory on the unix box which can be accessed by the path “~”

The command executed in the example below is -

/opt/sas/sasb ~/prog.sas -print ~/prog.lst -log ~/prog.log

SAS executes the sas program prog.sas located in my home directory (i.e. ~/) and sends any output to ~/prog.lst and the log to ~/prog.log.

Screes shot below shows the code in prog.sas located in my home directory…and then later the batch command I wanted to execute…

This screen shot shows the execution and the text getting printed to the terminal…

This screen shot shows the log printed to the ~/prog.log file…

The SAS example does not generate any SAS output…so ~/prog.st is not created for this instance.

SAS Macro to split a dataset into multiple datasets vertically with a common primary key

This macro splits a dataset to multiple datasets vertically with a common primary key. For eg, a dataset has 400 fields and 20,000 records. If we can split the dataset into two, with 200 fields and 20,000 records in each dataset with a common field like loan number as primary key would be helpful to load the details for analysis.

/*

Taking an example...

*/

data flags;

set sashelp.flags;

run;

/**

To be called like this...

%splitdsnverticallykey(dsn,varperdsn,keyvars=);

eg. %splitdsnverticallykey(sashelp.vtable,4,keyvars=memname libname);

Where -----------

dsn - libname.datasetname to be split

varperdsn - How many vars per dsn excluding the key variables

keyvars - specify the primary key variables

*/

%macro splitdsnverticallykey(dsn,varperdsn,keyvars=);

/* split the keyvars into individual macro var names*/

%let num=1;

%let keyvar=%scan(&keyvars,&num,' ');

%let keyvar&num=&keyvar;

%let keyvarstr=%str(%")&keyvar%str(%",);

%do %while(&keyvar ne );

%let num=%eval(&num + 1);

%let keyvar=%scan(&keyvars,&num,' ');

%let keyvar&num=&keyvar;

%if &keyvar ne  %then %let keyvarstr=&keyvarstr%str(%")&keyvar%str(%",);

%end;

%let numkeyvars=%eval(&num - 1);

%let keyvarstr=%substr(&keyvarstr,1,%length(&keyvarstr)-1);

data _null_;

/*Open the dataset and assign to handler*/

dsid=open("&dsn","i");

/*attrn with nvars gives u the count of variables */

numofvars=attrn(dsid,"nvars");

call symput('numofvars',numofvars-&numkeyvars);

/*identify total number of dsns would it fit excluding the key vars*/

totalnumdsns=ceil((numofvars-&numkeyvars)/&varperdsn);

call symput('totalnumdsns',totalnumdsns);

do i=1 to numofvars;

varname=trim(left(varname(dsid,i)));

if varname not in (&keyvarstr) then

do;

k+1;

/*Get the name of the variables into macro variables*/

call symput(compress('varname'||k),varname);

end;

end;

rc=close(dsid);

run;

%let totalnumdsns=&totalnumdsns;

%let numofvars=&numofvars;

%put The dataset &dsn with &numofvars of variables excluding variables {&keyvars} is split vertically into &totalnumdsns datasets;

/* name the datasets in sequence */

%let start=0;

%let end=0;

%do i=1 %to &totalnumdsns;

%let start=%eval((&i-1)*&varperdsn + 1);

%let end=%eval(&start + &varperdsn - 1);

%if &end ge &numofvars %then %let end=&numofvars;

%put start=&start end=&end;

data &dsn.&i; /*Note: There should be a blank after &dsn.&totalnumdsns*/

retain &keyvars;

set &dsn (keep=&keyvars

%do m=&start %to &end;

&&varname&m.

%end;);

run;

%end;

%mend splitdsnverticallykey;

options nosource;

%splitdsnverticallykey(flags,4,keyvars=title);

Log generated shows that 2 datasets flag1 and flag2 are created with 5 and 3 columns respectively…

flag1 dataset

flag2 dsn