Mean Absolute Percentage Error

Mean Absolute Percentage Error (MAPE) offers a relative measure of model accuracy, expressing errors as percentages of actual values, which is intuitive for comparing performance across different scales or variables. It's especially useful for forecasting tasks where understanding proportional errors matters. Lower MAPE values (closer to 0%) signal high accuracy, and it highlights relative error magnitudes, making it suitable for imbalanced or varying-scale data.

The formula for MAPE is:

$$\text{MAPE} = \frac{1}{n} \sum_{i=1}^n \left| \frac{y_i - \hat{y}_i}{y_i} \right| \times 100\%$$

Where$y_i$ is the actual value,$\hat{y}_i$ the prediction, and$n$ the number of points. This averages the absolute relative errors, multiplied by 100 for percentage form.

MAPE excels in contexts like time-series forecasting, but caution is needed with near-zero actuals to avoid division issues—consider weighted alternatives. It aids in evaluating proportional fit, ensuring insights align with real-world scales.

Its scale-independence makes it versatile, though it can be asymmetric for over/under-predictions. MAPE simplifies reporting of relative performance.