Absolute error and relative error
Machine translation
Learn
When you approximate a number, you no longer have the exact value. The exact value is the true one; the approximate value is the one you use instead. The difference between the two is the error.
The absolute error is the distance between the approximate value and the exact value. You subtract the smaller from the larger, so it is never negative: it does not tell you whether the approximate value is greater or less than the exact value. If the exact value is 3.47 and you round it to 3.5, the absolute error is 0.03.
E is the exact value and A the approximate one: the absolute error is the distance, 0.03 But the absolute error does not tell you whether the error is big. An error of 2 cm is small on a 40 cm table and big on a 10 cm pencil.
The same error of 2 cm: small on the table and big on the pencil The relative error compares the error with the size of the measurement: you divide the absolute error by the exact value. It is “part out of the whole”, as in percentages: the part is the error and the whole is the exact value. It is usually given as a percentage and has no unit. And the other way round: with the relative error and the exact value you find the absolute error, and with the absolute error and the relative error, the exact value.
Table Pencil Exact value 40 cm 10 cm Absolute error 2 cm 2 cm Relative error 5% 20% 2 ÷ 40 = 0.05, 5%; 2 ÷ 10 = 0.2, 20% To compare approximations of different measurements, look at the relative error: the better one is the one with the smaller relative error. With the same measurement, it is enough to compare the absolute errors. Careful: always divide by the exact value, and put both measurements in the same unit.
Measuring with an instrument also gives an approximate value: a ruler goes down to the millimetre and a thermometer to the degree. To find the error of a measurement you need to know the exact value.
Worked examples
Example 1
Ellie says her loaf weighs about 500 g, but it weighs exactly 400 g. What is the absolute error? And the relative error?
| Exact value | 400 g |
|---|---|
| Approximate value | 500 g |
| Absolute error | ? |
| Relative error | ? |
- The exact value is 400 g and the approximate value is 500 g. The absolute error is the distance: 500 − 400 = 100 g.
- The relative error is the absolute error divided by the exact value: 100 ÷ 400 = 0.25.
- 0.25 is : 25%.
The absolute error is 100 g and the relative error is 25%.
Watch out! Here many people divide by the approximate value: 100 ÷ 500 = 20%.
Example 2
Gus measures a 100 cm table and says 98 cm. Julie measures a 10 cm pencil and says 9 cm. Who approximates better?
| Gus | Julie | |
|---|---|---|
| Exact value | 100 cm | 10 cm |
| Approximate value | 98 cm | 9 cm |
| Absolute error | 2 cm | 1 cm |
| Relative error | 2% | 10% |
- The absolute errors are 2 cm and 1 cm. Julie's is smaller, but the measurements are not the same.
- Gus's relative error: 2 ÷ 100 = 0.02, 2%. Julie's relative error: 1 ÷ 10 = 0.1, 10%.
- Gus has the smaller relative error.
Gus approximates better: 2% compared with 10%.
Watch out! Here many people choose Julie because her absolute error, 1 cm, is smaller.
Example 3
Tony rounds 6.78 to the nearest tenth and writes 6.8. Mara truncates it to the tenth and writes 6.7. Who has the smaller absolute error?
- The exact value is 6.78. The rounded value, 6.8, is 6.8 − 6.78 = 0.02 away.
- The truncated value, 6.7, is 6.78 − 6.7 = 0.08 away.
- You always subtract the smaller from the larger: the absolute error is a distance.
Tony's: 0.02 compared with 0.08.
Watch out! Here many people subtract 6.7 − 6.78 and write −0.08: the absolute error is a distance and is not negative.
A challenge to think about
Use a ruler to measure a pencil or a book and round the measurement to the nearest centimetre. Work out the absolute error and the relative error of your rounding. What would happen to the relative error if the same absolute error were on something much longer?
Practise
Rounds of 10 questions. Choose a level:
Your grade in this topic:No grade yetDo a round at the Basic level to get a grade.
Next topic: Error bounds and error intervals when rounding →