POINT= option. Why does it need a variable?

[Oliver2]

Hello!

I had a quick look, but couldn't find any questions similar to this, so hopefully I'm not bringing up old questions.

What is the reason the the POINT= option only works when set to a set to equal a variable (eg the obsnum below), rather than just setting it to equal an integer?

eg why does this work

data OJOSData;

  obsnum=6;

  set grades point=obsnum;

  output;

  stop;

run;

but this doesn't?

data OJOSData;

  set grades point=6;

  output;

  stop;

run;

I was just wondering what the underlying reason is.

Many thanks

Oliver

PS the SAS University software is a godsend for testing things like this that puzzle me

[SASKiwi]

There are already options for reading a specific observation or range of observations. FIRSTOBS= for the first record to be read and OBS= for the last record to be read.

POINT= is for identifying observations by formula/calculation.

data want;

  set have (firstobs = 6 obs = 6);

run;

[Oliver2]

Right!  It's interesting isn't it how certain rules work in certain ways.

With POINT you HAVE to use a variable, but with firstobs you CANNOT use a variable.  At least I can't make it work...

data OJOSData;

  fobs=7;

  set grades (firstobs=fobs);

run;

Thanks SASKiwi, something for me to think about!

[SASKiwi]

That's right - refer to the documentation for the correct use of each option.

[gergely_batho]

obs= and firstobs= dataset options are designed to subset the data before it reaches the data step or a procedure. As they are data set options, they work almost in any circumstances. For example in a proc step. Obviously you don't have data step variables in a proc step.  obs= and  firstobs= can subset the observations very efficiently.

point= works only in a data step, it is an option of the set statement. It is designed to allow random/direct access to specific observations.

For example if you want to read the 10th, 20th and than again the 10th observation.