SMAPE (Symmetric Mean Absolute Percentage Error)

Definition

SMAPE is a forecast-accuracy metric — a variant of MAPE designed to handle the case where the actual value is very small or zero (where MAPE explodes). The error gets normalized by the average of forecast and actual, not just actual.

What it tells you

SMAPE is bounded between 0% and 200%, making it more robust on sparse data (e.g., low-occupancy days). Use SMAPE alongside MAPE when forecasting periods where some days have near-zero values — it won’t spike to misleading numbers.

How to track it

Computed after the period closes, like MAPE. Forecasting tools that publish accuracy usually publish both.

Where it fits

The companion to MAPE for forecast-quality measurement. Peaqplus publishes monthly MAPE / SMAPE / MAE on customer data to make forecast accuracy auditable.