Mean Absolute Error (MAE)

Mean Absolute Error, or MAE, is a standard metric for measuring forecast accuracy. It is the average of the absolute differences between predicted and observed values. For wave heights, an MAE of 0.8 ft means the forecast is, on average, within 0.8 feet of what actually happened. MAE is easier to interpret than root-mean-square error because it treats every foot of error equally and reports its result in the same units as the measurement. Rather than quoting a single MAE headline, Swell Intel records forecasts before the swell arrives and publishes the raw predicted-vs-observed pairs on the [forecast verification page](/accuracy) once the record has matured to a full month of wall-clock-recorded forecasts — so anyone can compute MAE, or any other error metric, for themselves.