Calculation problem in round up decimals C#

Question

1.00/5 (1 vote)

See more:

Hi all,
The question might sound a bit dumb, but I have been caught in a calculation scenario here.

There is a no. I divide it by a no. then again multiply it by same no. but I do not get exactly the same no. Decimal values are different. Can anyone please tell me a workaround of the problem.
Here is the code
Round1 and Round2 are having .00001 difference.

static void Main(string[] args)
{
decimal ab = new decimal();
ab = (decimal)4.444445;
decimal Round1 = Math.Round(ab,5);
decimal c = ab / 3;
decimal d = c * 3;
decimal Round2=Math.Round(d,5);
}

Posted 31-May-15 7:06am

Anant Beriwal

Add a Solution

Comments

Sergey Alexandrovich Kryukov 31-May-15 13:54pm

If course they are different. Rounding introduced rounding error. :-)
Any concerns about it?
—SA

Anant Beriwal 31-May-15 23:57pm

so any solution?? I need to round up. Can I provide some algo or hypothesis to it.

gggustafson 31-May-15 14:00pm

Math.Round is an abomination! It defers to "oddness" or "evenness" of a result. For that reason I wrote my own. See Known Color Palette Tool (http://www.codeproject.com/Articles/243610/The-Known-Colors-Palette-Tool-Revised-Again) in the section entitled (poorly) Extended Color class. The round function is in that section.

Sascha Lefèvre 31-May-15 14:16pm

Unrelated suggestion: You don't need to use "new" for primitive types. And you can tell the compiler a floating-point literal should be a decimal by appending an "m". The first two lines are therefore more elegantly written like so (single line):
decimal ab = 4.444445m;

Sergey Alexandrovich Kryukov 31-May-15 14:27pm

Sure.
—SA

1 solution

Add a Solution

Add your solution here

Treat my content as plain text, not as HTML

Preview 0

…

Existing Members

Sign in to your account

...or Join us

Download, Vote, Comment, Publish.

Your Email
Password
Forgot your password?

Your Email
This email is in use. Do you need your password?
Optional Password

I have read and agree to the Terms of Service and Privacy Policy
Please subscribe me to the CodeProject newsletters

When answering a question please:

Read the question carefully.
Understand that English isn't everyone's first language so be lenient of bad spelling and grammar.
If a question is poorly phrased then either ask for clarification, ignore it, or edit the question and fix the problem. Insults are not welcome.
Don't tell someone to read the manual. Chances are they have and don't get it. Provide an answer or move on to the next question.

Let's work to help developers, not make them feel stupid.

This content, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

Sergey Alexandrovich Kryukov · Accepted Answer · 2015-05-31T08:28:00

Solution 1

Please see my comment to the question.

I would advise the following: never ever round anything up. It just makes no sense, but can compromise the accuracy of calculations. There are cases when rounding error accumulate during calculation steps resulting in the error much greater than your required accuracy. Simply never round any calculations at all.

Instead, round up data only when you represent the final data in the string form, for example, when you show it on screen or store in a text file. At this step, you still don't need rounding itself, because it is already implemented in string formatting. In other words, reduce rounding problem to the formatting problem. Please see:
https://msdn.microsoft.com/en-us/library/system.double.tostring%28v=vs.110%29.aspx[^],
https://msdn.microsoft.com/en-us/library/kfsatb94%28v=vs.110%29.aspx[^],
https://msdn.microsoft.com/en-us/library/dwhawy9k(v=vs.110).aspx[^],
https://msdn.microsoft.com/en-us/library/0c899ak8(v=vs.110).aspx[^].

—SA

Posted 31-May-15 8:28am

Sergey Alexandrovich Kryukov

Comments

CHill60 31-May-15 14:30pm

All QA posts seem to be going through moderation queue. I wasn't quick enough to ok your solution and it's appeared twice - you might want to delete one copy

Sergey Alexandrovich Kryukov 31-May-15 14:41pm

Thank you very much. I will.
I don't think there is actual moderation. It looks like am old technical glitch, in a new form. When I refresh the page, it looks like non-posted text in the input box. Posting it results in the same situation repeated, or correct post, or double post. I failed to spot exact scenario, but it happened many times before, only without the word "pending".
—SA

CPallini 31-May-15 16:36pm

5.

Sergey Alexandrovich Kryukov 31-May-15 19:21pm

Thank you, Carlo.
—SA

Anant Beriwal 31-May-15 23:56pm

Sir,
The problem is, I have to round of. This is a real time scenario, the database will not hold anything beyond 5 decimal points. truncating the decimal set will cause even more difference with actual answer. I want to know, Can we do it someway correct.

Sergey Alexandrovich Kryukov 1-Jun-15 11:10am

It sounds like are just adding additional argument against rounding. Anyway, what does it mean "will not hold anything beyond 5 decimal points". What data type is that?
—SA

George Swan 1-Jun-15 1:44am

Try testing for equality in this way
bool isEqual = Math.Abs(Round1 - Round2) < 0.00002m;

Anant Beriwal 1-Jun-15 2:21am

this may work, as long as my use case is concerned. Going by a more generalize way, I was thinking will the rounding up cause issue in greater significance. specially in cases where money is involved.

Sergey Alexandrovich Kryukov 1-Jun-15 11:25am

The trouble is: you never properly explained where would you need rounding and what would you do with a different rounding method if you had it. What would be your scenario and why?
—SA

Anant Beriwal 5-Jan-16 5:29am

The actual problem is related to currency. We have a system which have data from one currency to another. the values are against 1 dollar. now each value more than 5 point decimal. Now when I sell a item, the user may pay in different currency. I convert it to USD (up to 5 points decimal) and save it at my database, also I will save rate of conversion for that day. Now when user asks for report at end of month/year, it will be in his own currency. now there this calculation gave errornous data. This is just a small version of that.
My database stores only upto 5 decimals. I cannot change design now

Sergey Alexandrovich Kryukov 5-Jan-16 10:35am

The subject of money type used to be a subject of heated discussion. :-)
You are talking about the aspect of money, where money should be represented as a floating-point value. But after the value exchange, the money can get into the system where fixed-cent calculations are required to be used. By the way, I've read about the criminal scheme based on money rounding.

I don't understand your "5 decimals". Do you mean strings with 5 characters? Such type should not be used to represent numeric values, money or any other. If your current database scheme is supposed to use such data type for money, the database may eventually fail to play its role completely.

—SA

Anant Beriwal 5-Jan-16 12:33pm

simple, it is a database field having length of decimal (18,5) (means 5 digits after decimal). so code picks it up at it is. problem is simple. when the transaction was done and now when the reports are generated, there is a difference of some dollars( there are more than 1000 transactions). I hit and try some other ways and find out that the way of calculation is the problem. The same issue lie with 3 decimal places too. See the code

static void Main(string[] args)
{
decimal ab = new decimal();
ab = (decimal)4.445;
decimal Round1 = Math.Round(ab,2);
decimal c = ab / 3;
decimal d = c * 3;
decimal Round2=Math.Round(d,2);
}

I have already closed the issue at my end. I changed the way how calculations will be done at time of actual transaction. (I removed the division part from equation). I am just curious how much bigger blunder this kind of thing can cause. Sorry but I am fond of full proof solutions :)

Sergey Alexandrovich Kryukov 5-Jan-16 13:32pm

I understand now, thank you. As I say, it can work, but this database design is really bad. For some calculations, it's good to have "fixed-precision" data type, which is essentially the integer type. Having a string used to represent numbers is not a good thing.
—SA