topic Re: use of sas base functions within PROC IML in SAS/IML Software and Matrix Computations

use of sas base functions within PROC IML

saibot — Wed, 02 Nov 2011 17:20:11 GMT

Hi all,

at first an example:

PROC IML;

a = {1 2 3} ; b = {1,2,3}; c = {1 2 3, 2 3 4, 3 4 5} ;

ssq_a = SSQ(a);

sum_b = SUM(b);

median_c = MEDIAN(c);

print a b c ssq_a sum_b median_c ;

std_a = STD(a) ;

std_c = STD(c) ;

print std_a std_c ;

QUIT ; *proc iml ;

My question: In the SAS Online Doc: http://support.sas.com/documentation/cdl/en/imlug/59656/HTML/default/viewer.htm#langref_sect315.htm there is a list of the "Base SAS Functions Accessible from SAS/IML". As you can see, if you run the example above, only some of the listed base functions work, like SUM, SSQ or MEDIAN, but not STD or CV. How can I use the sas base functions like STD, CV,... in IML?

Thank you in advance for every tip, best regards,

saibot

use of sas base functions within PROC IML

Rick_SAS — Wed, 02 Nov 2011 17:30:50 GMT

1) This issue is discussed and solved in Section 4.9 of my book, Statistical Programming with SAS/IML Software.

2) If you are using SAS IML 9.22 or beyond, you can use the VAR function, and of course STD=sqrt(VAR(..)). SAS 9.3 supports the STD function directly.

For more on this topic, see my blog: http://blogs.sas.com/content/iml/2011/06/08/calling-base-sas-functions-from-sasiml-how-to-handle-argument-lists/

Re: use of sas base functions within PROC IML

saibot — Thu, 03 Nov 2011 11:28:30 GMT

Thank you very much for your help, Rick. I tried the function GetArgList that you proposed in your blog and it worked fine.

I am very surprised that the use of such elementary functions in IML is so complicated. In some cases, for example for such simple functions like STD or CV, the programming "by hand" is easier to read than the three steps "GetArgList --> Creation of Macro Variable --> Use of SAS-Base function with macro variable". But as more base functions are available in higher SAS versions, it seems that SAS is working on that topic.

Thank you and best regards,

saibot