SAS it up

remove formats or labels

2009-07-26T18:06:00.000-07:00

This tip comes by way of sastips.com

sometimes

The following example of code removes all labels and formats for the dataset EXAMPLE in the directory referenced by the LIBNAME WORK:

    proc datasets lib=work nolist;
       modify example;
           attrib _all_ label=''; *Remove labels;
           format _all_;          *Remove formats;
       quit;
   run;

sas dates

2008-07-01T09:44:00.000-07:00

Dealing with SAS dates can be tricky, particularly if you're migrating your data between different analysis platforms (SAS in UNIX, R, JMP, etc.

In order to convert a date into the sasdate, you can use one of 2 methods, depending on whether the variable was read in as a character or numeric variable.

method 1: newsasdate=input(olddate, date11.);

method 2:newsasdate=input(put(olddate, Z6.) ddmmyy6.

magic code to transform R data with 'NA' for missing to SAS '.'

2008-03-17T17:26:00.000-07:00

array inc dlfib--dlcrp;
do i = 1 to 4;
if inc[i]='NA' then inc[i]=.;
end;
drop i;
run;

use "where" statements to print observations for selected critieria

2007-07-14T05:57:00.000-07:00

In this example I want to output data for specific ids I've selected in SAS. This turns out to be easier said than done...
The correct syntax for this is as follows.

proc print data=alldat2;
24725 where id=1 or id=2 or id=3;
24726 run;

NOTE: There were 1 observations read from the data set WORK.ALLDAT2.
WHERE id in (1, 2, 3);
NOTE: The PROCEDURE PRINT printed pages 4-14.
NOTE: PROCEDURE PRINT used (Total process time):
real time 22.44 seconds
cpu time 2.51 seconds

voila.

use "retain statement" to make a count variable

2007-06-24T06:51:00.000-07:00

**This code comes from UCLA statistical computing seminar
(link here to code)

proc sort data=kids out=sort_kids;
by famid;
run;
data retain1;
set sort_kids;
retain sumwt count; 
/*carry over the value from previous obs to next obs*/
by famid;
if first.famid then do;
/*at 1st obs of each family set sumwt and count = 0*/
  sumwt=0;
  count=0;
end;
sumwt = sumwt + wt;
count = count + 1;
meanwt = sumwt/count;
run;
proc print data=retain1;
var famid kidname wt sumwt count meanwt;
run;

code for transforming data from multivariate format to the univariate format

2007-05-30T14:22:00.000-07:00

data data;
infile 'c:\myfiles\testfile.dat';
input id y0 y1 y2 y3;
y=y0 trt=0; output;
y=y1 trt=1; output;
y=y2 trt=2; output;
drop y0-y2;
run;

quartiles (or x-tiles) for the lazy

2007-02-13T17:58:00.001-08:00

data composite;
set data.composite;
proc rank groups=4 out=ranked;
var srcalc srprot sriron srmagnes srphos srzinc srvitc srvitd packyrs bmi;
ranks rank_srcalc rank_srprot rank_sriron rank_srmagnes rank_srphos rank_srzinc rank_srvitc rank_srvitd rank_packyrs rank_bmi;
run;

create a categorical variable in one fell swoop

2007-02-11T13:08:00.000-08:00

categorize a variable with "lazy elegance"

if age ne . then agecat= int(iage/5-1);

make a good looking histogram

2007-02-11T13:00:00.000-08:00

More often than would be preferable, it becomes necessary to present data graphically. For a fancy ODS plot of a histogram, try the following:

ods pdf file ="%sysfunc (reverse(%sysfunc(substr(%sysfunc(reverse(%sysfunc(reverse(%scan(%sysfunc(reverse(%sysfunc(getoption(sysin)))),1,/))))),5)))).pdf"; goptions reset=global device=gif ftitle=swissb ftext=swissb htitle=2 htext=2; proc univariate data = alldat2 noprint; title "Histogram for pmh variables"; histogram pmh_dur3 pmh_dur4 pmh_dur5 pmh_dur8 drh_use lst_byirt/cfill=ligr normal cframe=liy barwidth=8 cv=black; inset mean std max min;

ods pdf close;

transpose data

2007-02-11T08:25:00.001-08:00

Transpose your data from the so-called "long format" to the ever popular and
occassionally useful "wide format"


proc transpose data=long2 out=widef prefix=faminc;
by famid;
id year;
var faminc;
run;

proc transpose data=long2 out=wides prefix=spend;
by famid;
id year;
var spend;
run;

data wide2;
merge  widef(drop=_name_) wides(drop=_name_);
by famid;
run;