Correct use of DO WHILE loop?

[genemroz]

Esteemed Advisers,

I know I'll be embarassed when the answer comes back on this question:

The goal of the attached code is to generate 10 random meteors within each of two volumes of sky. Each meteor must be at least 10 units in length.

The code produces 10 satisfactory meteors for Target A but none for Target B.

What is the correct way to formulate this DO LOOP to process both TARGET A and B in dataset TargetArea?

Thanks in advance,

Gene

Data TargetArea;
Input Target $ txmin txmax tymin tymax tzmin tzmax;
datalines;
A 0 5 0 5 70 120
B 0 10 0 10 70 120
run;
/* SAS macro from Wicklin for generating Random Numbers between a Min and Max */
%macro RandBetween(min, max);
   (&min + floor((1+&max-&min)*rand("uniform")))
%mend;
data Random_Meteors (keep= target meteorct1 meteorct2);
set work.targetarea;
meteorct1=0;
do while (meteorct2<10);
   TX_Start = %RandBetween(txmin, txmax);
   TY_Start = %RandBetween(tymin, tymax);
   TZ_Start = %RandBetween(tzmin, tzmax);
   TX_End = %RandBetween(txmin, txmax);
   TY_End = %RandBetween(tymin, tymax);
   TZ_End = %RandBetween(tzmin, tzmax);
   meteorct1+1;
   TrackLength=sqrt((TX_Start-TX_End)**2+(TY_Start-TY_End)**2+(TZ_Start-TZ_End)**2);
   /* if tracklength<10, generate another meteor until it finds one with tracklength >=10*/
   if tracklength<10 then continue;
   else;
   meteorct2+1;
   output;
end;
run;

[jimbarbour]

You have an error in your IF statement.  There is a semi-colon after the "else".  That means that Meteor2 is incremented unconditionally (every time) rather than checking for meteors of the proper size.  Delete the semi colon and re-run.

   if tracklength<10 then continue;
   else;
   meteorct2+1;

Jim

[genemroz]

Thanks for spotting this. Semi-colons tend become invisible after awhile.

[mkeintz]

You are using the summing statement

meteorct2+1;

which will retain the value of 10 achieved in the first observation as the starting value for the 2nd obs.  As a result the "do while (meteorct2<10);" loop will never be activated in the 2nd obs.

Use

meteorct2=sum(meteorct2,1);

instead.  It will not retain values across observations.

[jimbarbour]

Ah, good point @mkeintz 

I think, though, that in addition to the point raised by @mkeintz the erroneous semi-colon will need to be first removed before even the first observation is processed correctly.

Jim

[ChrisHemedinger]

You've received good findings from @jimbarbour and @mkeintz.  I'll just add that the DATA step debugger (available in SAS Enterprise Guide and SAS Studio) can be used to find logic/behavior errors like this!

[jimbarbour]

Thanks @ChrisHemedinger 

The interactive debugger is a truly beautiful feature of Enterprise Guide.  I really don't consider Enterprise Guide to have become excellent until version 7.1.3 (as I recall) came out which included the debugger.  Versions 5.1 up to 7.1.2 were good but not excellent.  Versions prior to 5.1, uh, well, they weren't so good.  🙂

Jim

[genemroz]

Thanks for this tip. I was not aware of the Data step debugger but will certainly make use of it in the future.

Gene