topic Where=(b1) in Statistical Procedures

Where=(b1)

Jabbawonga — Tue, 13 May 2014 15:42:00 GMT

Hello,

I'm working through some code in an old program written by someone else, and I came across the following lines:

Proc Reg ... (where=(b1))...

and

Proc Syslin ... (where=(b3))...

I've searched the documentation and don't see any explanation regarding "where" statements in Proc Reg or Syslin or what b1 and b3 are (they are not variables in the models). Can anybody help me understand what "where=(b1)" or "where=(b3)" mean?

Thanks.

Re: Where=(b1)

Reeza — Tue, 13 May 2014 15:46:17 GMT

Where statements are related to the input data set.

Its the equivalent of saying:

where b1 ne 0 or b1 ne .

or where b3 ne 0 or b1 ne .

Re: Where=(b1)

ballardw — Tue, 13 May 2014 15:58:07 GMT

You didn't post a whole line of code but I will bet that the (where= (b1)) immediately followed the input dataset name. This is one of many dataset options that are universal that allow selecting data, variables and renaming variables at time of execution.

Re: Where=(b1)

Amir — Tue, 13 May 2014 16:23:23 GMT

Hi,

Some documentation on "where=(...)" data set option:

SAS(R) 9.4 Data Set Options: Reference

In your example if b1 is numeric it should either be:

a) "." (missing)

b) "0" (zero)

c) Any other number (e.g. 12, -77, etc.)

(a) & (b) would be equivalent to False (observation would not be used) and (c) would be equivalent of True (observation will be used).

E.g. the second data step uses "where=(age)" to only process a valid age value:

data test1;

set sashelp.class;

if name=:'J' then

age=.;

run;

data test2;

set test1(where=(age));

run;

Regards,

Amir.