How it works
Pair entries by position, then square the prediction errors, average them and take a square root. Actual values 1, 3, 5 and predictions 1, 1, 7 have errors 0, −2, 2, giving RMSE √(8/3) and mean absolute error 4/3. Lists must have equal length and a common unit.
Inputs and units
| Input | Example value | What to enter |
|---|---|---|
| Actual values | 1 3 5 | Separate numbers with spaces or semicolons; use a decimal point. |
| Predicted values | 1 1 7 | Separate numbers with spaces or semicolons; use a decimal point. |
Worked example
Show example inputs and results
This example is generated by the same calculation engine using the starting values in the form. Your own inputs produce your own result.
- Actual values
- 1 3 5
- Predicted values
- 1 1 7
Complete example results
- Root mean squared error
- 1.63299316
- Mean absolute error
- 1.33333333
Reading the result
Keep the list to one kind of measurement. Outliers can affect the mean; sample and population formulas use different denominators. The tool applies its stated method and does not establish the quality of your data.
Questions and answers
Can I use the example values for my own calculation?
The example values illustrate the calculation. Replace every input with your own information, especially dates, rates and the selected operation.
Why can RMSE be larger than mean absolute error?
Squaring gives larger errors more influence before the square root is taken. Compare models on the same observations and measurement scale; the metric alone does not establish a useful forecast.
Why are some results rounded?
Results display up to 8 decimal places. Calculations use the unrounded values; very small values use scientific notation. JavaScript floating-point arithmetic can introduce tiny rounding differences.
Are my inputs sent to a server?
The calculator processes inputs in your browser. Its code does not send your values to a server.