What the accuracy scores mean
Error, simple method, bias, MASE and range hit rate, in plain words: what each one tells you and what's good.
How they're measured: we hide your most recent sales (for example the last 14 days), forecast them without looking, then compare with what really sold.
Error (WAPE): how far the forecast was from actual sales, overall. 13% means it missed by about 13 units for every 100 sold. Lower is better.
Simple method (baseline): the error you'd get by repeating the same weekday last week. The forecast is only worth using when its error is lower; if not, we use the simple method and say so.
Bias: whether the forecast leans high or low. Above 0 means it forecast more than sold, below 0 less. Close to 0 is best.
MASE: the error compared with simply repeating the last period. Below 1 beats it: 0.81 means about 19% less error.
Range hit rate: how often actual sales fell between Low and High. For an 80% range, about 80% is right; much lower means the range is too narrow for this data.
How to read your forecast → · Why the simple method →