CODEWITHSUNDEEP
javascriptaverage
SWeko
SWeko

Reputation: 30902

Calculate average of two numbers with the same precision

I need to calculate the middle point (average) between two real numbers in JavaScript. The range of the numbers can vary widely, generally between 10000 and 0.0001.

The naive approach

(parseFloat(first) + parseFloat(second)) / 2

gives me unwanted precission errors, i.e.

(1.1 + 0.1) / 2 = 0.6000000000000001

How can I ensure that the result does not have extra decimal spaces? I guess, since there are two and only two inputs, that the result will need to have maximum one more decimal place than the inputs. So, I need:

 1000 and 3000 to return 2000 (without decimal spaces)
 1234.5678 and 2468.2468 to return 1851.4073
 0.001 and 0.0001 to return 0.00055
 10000 and 0.0001 to return 5000.00005
 0.1 and 1.1 to return 0.6

To clarify: I know all about precision errors and why this happens. What I need is a simple workaround, and I have not been able to find a previous solution on SO.

Upvotes: 0

Views: 6783

Answers (5)

Graham Walters
Graham Walters

Reputation: 2064

You'll want to use the function toFixed(). Here's what your code could look like ((1.1 + 0.1) / 2).toFixed(4);. And 4 is the number of decimal spaces.

Here's a sample

var num1 = "123.456";
var num2 = "456.1235";
if (num1.split('.')[1].length > num2.split('.')[1].length)
  var x = ((parseFloat(num1) + parseFloat(num2)) / 2).toFixed(num1.split('.')[1].length+1);
else
  var x = ((parseFloat(num1) + parseFloat(num2)) / 2).toFixed(num2.split('.')[1].length+1);

And remember inputs are strings, so that's why num1 and num2 are strings.

Update: Here's the correct if statements for your desired results.

var num1 = "123.456";
var num2 = "456.1235";
if (num1.split('.').length == 1) {
  if (num2.split('.').length == 1)
    var x = ((parseFloat(num1) + parseFloat(num2)) / 2).toFixed(0);
  else
    var x = ((parseFloat(num1) + parseFloat(num2)) / 2).toFixed(num2.split('.')[1].length);
} else if (num2.split('.').length == 1)
  var x = ((parseFloat(num1) + parseFloat(num2)) / 2).toFixed(num2.split('.')[1].length);
else if (num1.split('.')[1].length == num2.split('.')[1].length)
  var x = ((parseFloat(num1) + parseFloat(num2)) / 2).toFixed(num2.split('.')[1].length);
else if (num1.split('.')[1].length > num2.split('.')[1].length)
  var x = ((parseFloat(num1) + parseFloat(num2)) / 2).toFixed(num1.split('.')[1].length+1);
else if (num1.split('.')[1].length < num2.split('.')[1].length)
  var x = ((parseFloat(num1) + parseFloat(num2)) / 2).toFixed(num2.split('.')[1].length+1);

EDIT I had to change the order it checked for numbers with no decimal space. Here's a jsFiddle I created to help with the testing: link

Upvotes: 3

kennebec
kennebec

Reputation: 104780

You can use the 'precision' of the numbers in use to judge the appropriate precision of the return value.

This example averages any number of numbers to the greatest precision of its arguments.

Math.average= function(){
    var a= arguments, L= a.length, i= 0,
    total= 0, next, prec= [];
    if(L== 1) return +a[0];
    while(i<L){
        next= Number(a[i++]);
        total+= next;
        if(next!== Math.floor(next)){
            prec.push(String(next).split('.')[1].length);
        }
    }
    prec= Math.max.apply(Math, prec)+1;
    return +((total/L).toFixed(prec));
    // returns number after precision adjusted
}

/*

tests:
var n1= ;
Math.average(1.1, .01);
0.555

var n1= 1.0111, n2= .01;
Math.average(1.0111,  .01);
0.51055

var n1= 1.1, n2= .01, n3= 1, n4= 1.025;
Math.average(n1, n2, n3, n4)
0.7838

*/

Upvotes: 1

Philipp
Philipp

Reputation: 69663

You can use Number.toPrecision(digits) to do that.

Upvotes: 0

Philipp
Philipp

Reputation: 69663

Numbers in Javascript have the toFixed(n) method which removes all but n decimal places. This, however, isn't very helpful when you don't know if you have a very large or very small number. But what you can do is first take Math.floor() of the logarithm to base 10 of the number to get the number of decimal places in front of the dot, divide the number by 10^places, perform toFixed(n) where n is the maximum number of digits, and then multiply it again with 10^places.

Upvotes: 2

Denys S&#233;guret
Denys S&#233;guret

Reputation: 382170

Numbers in javascript are all double precision floats as defined by IEEE754. This means there isn't a dot position defined in base 10 and that you'll always find those cases where the precision doesn't seem to be what you want.

The solution is either

  • to not use js numbers to do your computation. This solution is heavy and generally useless.
  • to format your numbers into strings when you need to display them (and only at that time). You may use to Fixed or toPrecision.

Useful read : What Every Computer Scientist Should Know About Floating-Point Arithmetic

Upvotes: 1

Related Questions