Calculator Input
Example Data Table
| Row | Actual | Predicted | Absolute Error | APE % |
|---|---|---|---|---|
| 1 | 120 | 115 | 5 | 4.17% |
| 2 | 150 | 160 | 10 | 6.67% |
| 3 | 180 | 175 | 5 | 2.78% |
| 4 | 200 | 195 | 5 | 2.50% |
| 5 | 210 | 205 | 5 | 2.38% |
| Example MAPE | 3.70% | |||
This sample shows how small row-level differences combine into an overall percentage error score.
Formula Used
MAPE (%) = (100 / n) × Σ |(Actual − Predicted) / Actual|
This page uses the absolute value of the actual observation in the denominator for percentage scaling. That keeps the percentage error positive and easier to interpret across datasets with mixed signs.
Supporting diagnostics:
- MAE = average absolute error.
- RMSE = square root of average squared error.
- MPE = average signed percentage error.
- sMAPE = symmetric percentage error using actual and predicted magnitudes.
- WAPE = total absolute error divided by total actual magnitude.
How to Use This Calculator
- Enter a model name to label the report.
- Paste actual values into the first field.
- Paste predicted values into the second field.
- Select the separator mode if your data follows a fixed pattern.
- Choose how zero actual values should be handled.
- Set decimal precision and enable the per-row table if needed.
- Press Calculate MAPE to show the result above the form.
- Use the CSV or PDF buttons to export the generated report.
8 FAQs
1. What does MAPE measure?
MAPE measures the average absolute percentage difference between actual and predicted values. It helps compare forecast quality across models when stakeholders prefer percentage-based interpretation.
2. Why are zero actual values a problem?
MAPE divides by the actual value. When the actual value is zero, the percentage calculation becomes undefined. This calculator lets you exclude those rows, stop the process, or use epsilon.
3. Is a lower MAPE always better?
Yes, lower MAPE generally means predictions are closer to actual values. Still, you should review MAE, RMSE, and bias too, because MAPE alone can hide model behavior.
4. When should I avoid MAPE?
Avoid relying only on MAPE when your dataset has many zeros, tiny actual values, or strong sign changes. In those cases, sMAPE, WAPE, MAE, or RMSE may be more stable.
5. What is the difference between MAPE and sMAPE?
MAPE scales error by the actual value only. sMAPE uses both actual and predicted magnitudes, making it less sensitive when actual values are very small.
6. Can I use negative values?
Yes. The calculator uses the magnitude of the actual value for percentage scaling. That keeps the percentage interpretation readable even when some observations are negative.
7. What does forecast accuracy mean here?
This page defines forecast accuracy as 100 minus MAPE, floored at zero. It is a simple reporting aid, not a universal machine learning standard.
8. What is included in the exported files?
The exports include the model label, key summary metrics, zero-handling note, and row-level details when available. This makes the results easier to document and share.