Rounding

[JMK NI]

Well knock me down! It's 6:30AM, I haven't had coffee yet!

[PaulowniaK]

OK, bad example...

123 would be 120.
125 would be 120.
126 would be 130.

etc. etc.

The point not being the number of significant digits you end up with, but the fact that the rounding is happening at 5 and 6 rather than the more usual 4 and 5.

Almost, but not quite, entirely unlike... me...

[DaveAuld]

Well it's now 6:43am, my first meeting with my supervisory team was an hour ago and I have already spoken with my boss onshore, so the day is well underway. Only another 14 or 15 hours still to go. :doh: Feeling :zzz: already.

Dave Find Me On:Web|Facebook|Twitter|LinkedIn Folding Stats: Team CodeProject

[PaulowniaK]

My understanding of rounding to N significant figure is like this:

1234 to 3 significant figures = 1230
1235 to 3 significant figures = 1240
1236 to 3 significant figures = 1240

The trouble is, I need

1234 to 3 significant figures = 1230
1235 to 3 significant figures = 1230
1236 to 3 significant figures = 1240

The point is, rather than rounding up or down depending on whether the N+1th digit is 4 or 5, but whether it is 5 or 6.

Almost, but not quite, entirely unlike... me...

[DaveAuld]

Sounds like you want to do Round Half Down + Significant then.

Dave Find Me On:Web|Facebook|Twitter|LinkedIn Folding Stats: Team CodeProject

[DaveAuld]

Round half down + Significant.

Dave Find Me On:Web|Facebook|Twitter|LinkedIn Folding Stats: Team CodeProject

[Agent__007]

How about "N Significant Half-Down Rounding[^]"?

Whether I think I can, or think I can't, I am always bloody right!

[Jorgen Andersson]

So depending on how you want to treat negative numbers you want either "Significant digits round half down" or "Significant digits round half towards zero".

Wrong is evil and must be defeated. - Jeff Ello[^]

[Jorgen Andersson]

See my other answer, or Daves.

Wrong is evil and must be defeated. - Jeff Ello[^]

[PaulowniaK]

Cheers. I think "round half down" is the term I needed. :thumbsup:

Almost, but not quite, entirely unlike... me...

[PaulowniaK]

All values are positive, due to the nature of the application. "Round half down" was the term I was looking for. Thanks for your help. :cool:

Almost, but not quite, entirely unlike... me...

[PaulowniaK]

Thanks, "Round half down" was the term I was looking for. :)

Almost, but not quite, entirely unlike... me...

[Duncan Edwards Jones]

...or rounding to the most significant digit. (I'd suggest that we call this "politicians rounding")

[BillWoodruff]

JMK-NI wrote:

Disregard, I'm an idiot

Well, JMK-NI, don't worry too much: you are in the place for idiocy. It's kind of like a "contagious skin rash," here, on the Lounge, it breaks out in different people, and makes different people scratch in different places, while others scratch their heads. cheers, Bill

“I have diligently numbered the days of pure and genuine happiness which have fallen to my lot: They amount to 14.” Abd-Ar Rahman III, Caliph of Cordoba, circa 950CE.

[Michael Pauli]

The 1.5 -> 1 is a little strange. Except for that you can use

double MagicFormula(double x)
{
var d = Math.Log10(x);
var f = Math.Pow(10.0, d);
var res = Math.Round(x / f, 0) * f;
var res2 = Math.Round(res, 0);
return res2;
}

Michael Pauli

[Gary Wheeler]

I think the word you are looking for is 'wrong'.

Software Zen: delete this;

[newton saber]

Kind of like ceiling / floor, but that's for decimal. http://en.wikipedia.org/wiki/Floor_and_ceiling_functions[^]