Forecasters¶

HAFO provides forecasting algorithms that transform historical entity data into future predictions.

Available Forecasters¶

Historical Shift¶

The primary forecasting algorithm that shifts historical data forward in time.

• Fetches hourly statistics from the recorder
• Shifts timestamps forward by N days
• Cycles the pattern to fill the forecast horizon

Choosing a Forecaster¶

Currently, HAFO offers one forecasting algorithm. Future versions may include additional algorithms such as:

• Weighted moving averages
• Seasonal decomposition
• Machine learning models

How Forecasts Work¶

All forecasters in HAFO follow a similar pattern:

1. Data Collection: Fetch historical data from the recorder
2. Transformation: Apply the forecasting algorithm
3. Horizon Filling: Extend the forecast to cover the desired time range
4. Output: Provide forecast as sensor attributes

The forecast is refreshed hourly to incorporate new data.