SAS slop on the internet

[BruceBrad]

Where do pages like https://answerpoint.blog/sas-not-equal-operator-guide come from? I could see two major errors before I gave up. Is this AI or human-written junk?

( "<>" does not mean 'not equal' in SAS data step code (though it does in SQL), and  " . ne 100" does not return false). 

[PaigeMiller]

The article tells you it comes from someone named Peter Schneider.

[Kurt_Bremser]

---

@PaigeMiller wrote:

The article tells you it comes from someone named Peter Schneider.

---

It links to the youtube channel of Schneider, but I find absolutely no reference to an author within the text.

The whole website shows no impressum or similar, making it look very suspicious.

[data_null__]

It means MAX in a data step.  SAS even tells you so.  The MIN and MAX operators are somewhat obscure.

58         data _null_;
59            x = 4 <> 3;
NOTE: The "<>" operator is interpreted as "MAX".
60            y = 4 max 3;
61            z = . ne 100;
62            put _all_;
63            run;

x=4 y=4 z=1 _ERROR_=0 _N_=1

[BruceBrad]

Yes, they've put that warning in the log because in other languages (including SAS SQL) it means 'not equals'.

Makes me think it might be AI written. Taking wrong information from closely related documents is the sort of thing that LLMs do. 

[ChrisHemedinger]

The video that appears to be the "source" is 5 years old, but the article is new. I agree it must be AI generated but why make this? I don't know.

[quickbluefish]

With pearls of wisdom like:

While SAS excels at handling vast datasets, its true analytical power is often unleashed through precise control and conditional logic.

...who can resist?  The sad thing is that since not anyone in their right mind would read an article even 1/10th that long whose sole topic was the concept of "not equal to" in a high level language, the only thing that will actually ingest it is other LLMs.  Slop begets slop.  

[Amir]

Hi _all_,

Reading all the posts about what the "<>" operator means, it did bring about memories for me where I did use it to mean not-equal-to in data step code.

Checking the documentation, I did find a page about operators with a section on MIN and MAX Operators that includes a table where the 2nd foot note says 'In a WHERE expression, the symbol representation <> is interpreted as “not equal to”.'

So, for example, the following code:

data want1;
  set sashelp.class;
  
  where age <> 15;
run;

yields the following log:

 73         data want1;
 74           set sashelp.class;
 75         
 76           where age <> 15;
 NOTE: The "<>" operator is interpreted as "not equals".
 77         run;
 
 NOTE: There were 15 observations read from the data set SASHELP.CLASS.
       WHERE age not = 15;
 NOTE: The data set WORK.WANT1 has 15 observations and 5 variables.

The "NOTE:" about the "<>" operator clarifies how it is being interpreted; I don't recall it ever misinterpreting my intention, but ymmv. The 2nd "NOTE:" helpfully spells out "WHERE age not = 15", as per usual SAS behaviour.

Comparing character data is no different, e.g.:

data want2;
  set sashelp.class;
  
  where name <>: 'J';
run;

 has the following log:

 71         data want2;
 72           set sashelp.class;
 73         
 74           where name <>: 'J';
 NOTE: The "<>" operator is interpreted as "not equals".
 75         run;
 
 NOTE: There were 12 observations read from the data set SASHELP.CLASS.
       WHERE name not =: 'J';
 NOTE: The data set WORK.WANT2 has 12 observations and 5 variables.

Lastly, both meanings can appear in the same data step:

data want3;
  set sashelp.class;
  
  where name <>: 'J';
  
  weight2 = weight <> 100;
run;

gives:

 71         data want3;
 72           set sashelp.class;
 73         
 74           where name <>: 'J';
 NOTE: The "<>" operator is interpreted as "not equals".
 75         
 76           weight2 = weight <> 100;
 NOTE: The "<>" operator is interpreted as "MAX".
 77         run;
 
 NOTE: There were 12 observations read from the data set SASHELP.CLASS.
       WHERE name not =: 'J';
 NOTE: The data set WORK.WANT3 has 12 observations and 6 variables.

So, in short, I wasn't just dreaming! 😆

Thanks & kind regards,

Amir.

[Tom]

That is because the WHERE is not run "IN" the data step, put as a filter on the data before it makes it to the data step.