topic Re: proc iml issue with concatenation in SAS/IML Software and Matrix Computations

proc iml issue with concatenation

ss59 — Tue, 19 Dec 2017 22:40:46 GMT

I have the following code that perfectly works:

proc iml;
a = {1,2,3,4,0};
b = {1 1, 1 2, 2 3,3 1, 2 2};
c = a||b;
d = full(c);
print a,b,c,d;
quit;

Now, I'm trying to the same thing with one row of my data but getting errors. My code is:

proc iml;
  use test(keep = TR_:);
  read all var _num_ into x [colname=vn];
  b = t(scan(compress(vn,'S'),2,'t_') // scan(compress(vn,'S'),3,'t_'));
  a = t(x);
  c = a||b;
  d = full(c);
  print x, a, b, c, d;
quit;

I can't somehow get the matrix 'c' correctly, it is throwing some error. Can anyone please point out where I'm going wrong?

The test file is following:

Year	Segment1	TR_S6_S5	TR_S7_S5	TR_S6_S7	TR_S7_S6	TR_S7_S8	TR_S6_S8	TR_S3_S1	TR_S2_S1	TR_S2_S3	TR_S3_S2	TR_S3_S4	TR_S2_S4	TR_S6_S2	TR_S7_S3	TR_S5_S1	TR_S8_S4	TR_S2_S6	TR_S3_S7
2017	ABC	0.001	0.3	0	0	0.02	0	0	0	0	0	0	0	0	0	0	0	0	0

Re: proc iml issue with concatenation

ballardw — Tue, 19 Dec 2017 22:51:33 GMT

Paste the code and messages from the Log into a code box opened with the forum {I} menu icon.

Re: proc iml issue with concatenation

ss59 — Tue, 19 Dec 2017 23:05:48 GMT

Got it. The reason was actually the fact that I didn't do num(scan(compress(vn,'S'),2,'t_')) so b was character matrix in disguise of numeric.

Following code works:

proc iml;
  use test(keep = TR_:);
  read all var _num_ into x [colname=vn];
  b = t(num(scan(compress(vn,'S'),2,'t_')) // num(scan(compress(vn,'S'),3,'t_')));
  a = t(x);
  c = a||b;
  d = full(c);
  print x, a, b, c, d;
quit;

Re: proc iml issue with concatenation

PeterClemmensen — Fri, 22 Dec 2017 09:11:00 GMT

@ss59 glad you found your answer 🙂

1 small thing that itches every IML programmer about this code is that you should always remember to close your data sets when you are done reading from them. So add a

close test;

statement after your read statement.

For a (5) reason(s) why, read Five Reasons to CLOSE Your Data Sets