PREDICTION ERROR
RMSE Calculator
Compare actual values with model predictions to calculate root mean squared error (RMSE), mean squared error (MSE) and mean absolute error (MAE). Results are calculated from the pairs you provide; this page does not fit a model.
YOUR DATA
Enter actual and predicted values
Paste paired numeric columns or upload a CSV/TSV file. Use a period as the decimal mark. Up to 5 MB and 5,000 observations are accepted.
The file is read locally in your browser.
One pair per row; actual and predicted values must be finite numbers.
YOUR RESULTS
Prediction error
Errors by observation
Error = predicted − actual; absolute error and squared error are also shown.
| Row | Actual | Predicted | Error | Absolute error | Squared error |
|---|
Formula and calculation method
For n paired observations, errorᵢ = predictedᵢ − actualᵢ. MSE = Σ(errorᵢ²) / n. RMSE = √MSE. MAE = Σ|errorᵢ| / n. Every row is included; the calculator reports invalid rows rather than silently ignoring them.
How to interpret RMSE and MAE
Both metrics describe error magnitude in the units of the target. RMSE gives more weight to large misses, while MAE treats each absolute miss linearly. A value is meaningful only relative to the scale, distribution and decision context of the target. Compare models on the same observations and avoid using these metrics alone to claim future performance.
Limitations
- Actual and predicted values must be aligned row by row and use the same units.
- RMSE and MAE do not show whether errors are systematically high or low; inspect the signed row errors.
- The calculator does not account for sampling, uncertainty, weighting, or model training data.
- A small average error does not guarantee safe performance for every subgroup or use case.