Problem with the round function! Please help

[deleted_user]

I tried the following code:

data testo; orig=0.54999999999999; output; Run;
Data testo; Set testo; rnd= round(orig,0.1); Run;

Could you please explain me what happened?? It should be rounded to 0.5 and not 0.6!!

Thanks for your help,
Regards,

[Cynthia_sas]

Hi:
You might look at the documentation on ROUND, the ROUNDE and the ROUNDZ functions...they operate a bit differently when rounding. According to the doc they operate this way:
"Computing Results When the Value Is Halfway between Multiples of the Rounding Unit
When the value to be rounded is approximately halfway between two multiples of the rounding unit, the ROUND function rounds up the absolute value and restores the original sign.
<<skip long program example>>
The ROUND function is the same as the ROUNDE function except that when the first argument is halfway between the two nearest multiples of the second argument, ROUNDE returns an even multiple. ROUND returns the multiple with the larger absolute value.

The ROUNDZ function returns a multiple of the rounding unit without trying to make the result match the result that is computed with decimal arithmetic."

You might experiment and see whether ROUNDZ gives you the kind of rounding your want.

cynthia

[deleted_user]

Thank you very much for your great help!

Best regards,

[hpl104]

Hi Cynthia,

I tried ROUND, ROUNDE, and ROUNDZ functions and my results still do not turn out correctly. I welcome any suggestion(s) you may have for me.

Thanks,

Hui

[Doc_Duke]

Hui,

When I did roundz, I got the answer Cynthia described.  (SAS 9.3 on Windows).  Maybe you need to show your log.

Doc Muhlbaier

Duke

data testo; orig=0.54999999999999; output; Run;

Data testo; Set testo;

rnd= round(orig,0.1);

rnde= rounde(orig,0.1);

rndz= roundz(orig,0.1);

FILE LOG;  PUT orig rnd rnde rndz;

Run;

---------

16   Data testo; Set testo;

17   rnd= round(orig,0.1);

18   rnde= rounde(orig,0.1);

19   rndz= roundz(orig,0.1);

20   FILE LOG;  PUT orig rnd rnde rndz;

21   Run;

0.55 0.6 0.6 0.5

[hpl104]

Thanks Dr. Muhlbaier. Those functions work on single numbers but not when multiplication/division is involved. Do you happen to know the equivalent SAS syntax/functon for ROUND(+B50/J50,0)) in MS Excel? I assume that the plus(+)  sign indicate a positive number and not a ceiling function but I could be wrong. Below is part of my syntax:

result(i)= round(numerator(i)/denominator(i));

or should I use result(i)= round((ceiling(numerator(i)))/(denominator(i))) instead?

I wonder if there is anything missing in my code .  . .

Thanks again,

Hui

PS: I have to convert various Excel pivot tables into SAS datasets and my results differ.

[Doc_Duke]

Unfortunately, I don't know enough about Excel to help.

[Reeza]

AFAIK and from a very quick test in Excel the + doesn't do anything.

I've seen people use it as a shortcut instead of typing equals because then excel treats it as a number/formula, and then the round was possibly introduced later on and then doesn't make sense.
Or apparently and old habit from a Lotus user http://www.excelbanter.com/showthread.php?t=178349

From a quick look at how the function works my guess would be you're looking for

round(numerator(i)/denominator(i), 1);

[hpl104]

Thanks alot Reeza. Regardless of how I write the ROUND function, I find that rounding seems to 'operate' differently in SAS and Excel. For instance 46.54421 is rounded to 47 in Excel and 46 in SAS and 46.4645 is rounded to 46 in Excel and 47 in SAS. I will check the link out.

[Reeza]

No it doesn't...46.544 rounds to 47 and 46.4645 rounds to 46 as expected in both SAS 9.2 and 9.3

I couldn't find a consistent way to get your results either, so you may want to post the code and the numbers you're trying to replicate.

2119  data _null_;

2120  x= round(46.54421, 1);

2121  y= round(46.4645, 1);

2122

2123  put x y;

2124  run;

47 46

[hpl104]

Not when I do divisions. 46.54421 and 46.4645 are elements of result(i).

result(i)= round(numerator(i)/denominator(i))

Thanks, Reeza.

[Reeza]

What are numerator(i) and denominator(i)? And what format is applied to the numbers

I still can't get those numbers you're getting, so there's something missing somewhere:

data test;

x= 46.54;

y= 46.46;

do i =1 to 5;

    numerator1=4654*i;

    numerator2=4646*i;

    denominator=i*100;

    unround1=numerator1/denominator;

    unround2=numerator2/denominator;

    round1=round(numerator1/denominator);

    round2=round(numerator2/denominator);

    output;

end;

run;

proc print data=test;

run;

[hpl104]

demoninator = 49.2333333333333333

numerator = 1.057775751

Thanks again, Reeza

1.057775751

1.057775751

[Reeza]

It works fine for me still, rounds to 47 as expected, there's something else in code that's causing this issue, not the round function, provide the full code for the data step.

183  data _null_;

184      x= round(49.23333333/1.0577775751, 1);

185      x2=49.233333/1.0577775751;

186      put x x2;

187  run;

47 46.54412625

NOTE: DATA statement used (Total process time):

      real time           0.00 seconds

      cpu time