Rounding with DecimalFormat in Java

Question

Let's look at the following statements in Java.

System.out.println(new DecimalFormat("0").format(2.4)); //returns 2

System.out.println(new DecimalFormat("0").format(2.5)); //returns 2  <---Concentrate here
System.out.println(Math.round(2.5));                    //returns 3

System.out.println(new DecimalFormat("0").format(2.6)); //returns 3
System.out.println(new DecimalFormat("0").format(3.5)); //returns 4

In the above statements, all other cases are obvious except the following.

System.out.println(new DecimalFormat("0").format(2.5));

It should return 3 but it returns 2. How?

Answer 1

The default rounding mode of DecimalFormat is RoundingMode.HALF_EVEN. This means that it rounds up, or rounds down if the number is nearer to the next neighbour. When the number is exactly between two neighbours (in your case, 2 and 3), it rounds to the nearest even number (in your case, 2).

As you can see, when you tried it with 3.5, it rounded to 4.

If you want the more "intuitive" behaviour, known as “schoolhouse rounding” as it is often taught to children, use RoundingMode.HALF_UP.

setRoundingMode(RoundingMode.HALF_UP)

If the number is exactly between two neighbours, HALF_UP will always round upwards.

Answer 2

This is intentional behavior. From the documentation:

Rounding

DecimalFormat uses half-even rounding (see ROUND_HALF_EVEN) for formatting.

About ROUND_HALF_EVEN:

Rounding mode to round towards the "nearest neighbor" unless both neighbors are equidistant, in which case, round towards the even neighbor. Behaves as for ROUND_HALF_UP if the digit to the left of the discarded fraction is odd; behaves as for ROUND_HALF_DOWN if it's even. Note that this is the rounding mode that minimizes cumulative error when applied repeatedly over a sequence of calculations.

This is also known as banker's rounding.

Math.Round on the other hand uses the following formula, which is "normal" rounding:

(long)Math.floor(a + 0.5d)