Paste any ordered series — sales, signups, call volumes, temperatures — and every average you can pick (simple, exponential, weighted) is scored against the data's own history before it dares to predict. The drift forecast comes with error bands sized from the backtest, so you see exactly how much the “forecast” is worth.
Rows may be “value” or “label, value”. Lines starting with # are ignored.
| step | forecast | low | high |
|---|
| method | MAPE | mean abs err | lag |
|---|