--forecast¶
Forecast one column over time with Holt-Winters: the level, the trend and a repeating season, with 80% and 95% ranges. The same model as the chart's Forecast; see Trend and forecast.
octa --forecast monthly_sales.csv --forecast-x month --forecast-y sales
octa --forecast visits.csv --forecast-x day --forecast-y visits --forecast-periods 14 --forecast-season 7
Flags¶
| Flag | Required? | Description |
|---|---|---|
--forecast |
yes | The file. |
--forecast-x |
yes | Time column: dates, date-times or numbers, evenly spaced. |
--forecast-y |
yes | Column of numbers to forecast. |
--forecast-periods |
no | How many periods ahead. Default 12. |
--forecast-season |
no | Season length in points. Default: read from the spacing. |
Output¶
For samples/features/monthly_sales.csv (the first two of the twelve rows):
x forecast lo80 hi80 lo95 hi95
2026-01-01 148.435 146.325 150.545 145.208 151.662
2026-02-01 155.736 152.752 158.72 151.172 160.3
The season length used goes to stderr (none when there was not enough
history for one). The x values must be evenly spaced; bucket uneven data
first, for example with --resample.
See also¶
- MCP:
forecast