TD_HOLT_WINTERS_FORECASTER is a time series forecasting method that uses a combination of level, trend, and seasonality components to make predictions. This method is exponential smoothing method, and is an extension of simple exponential smoothing. The function uses triple exponential smoothing on a forecast model with seasonal data or double exponential smoothing for non-seasonal data, depending on the parameters.
The seasonal aspect means the data is repetitive over some period. The level, trend, and seasonality components are used for the calculations. The level and seasonality components are additive or multiplicative. The trend remains additive.
- Level: The average value of the time series.
- Trend: The rate of change of the time series.
- Seasonality: The periodic fluctuations or seasonal patterns in the time series.
The alpha, beta, and gamma parameters of the function coincide with the level, trend, and seasonality, respectively, to control the main smoothing components. When values are not supplied for these parameters, their values are calculated, and available in the secondary result set.
Exponential smoothing is a weighted average of past observations, where the weights decrease exponentially as the observations get older. The weights are determined by the smoothing parameters level, trend, and seasonality, which are values between 0 and 1.
Once the components are estimated, TD_HOLT_WINTERS_FORECASTER uses them to forecast future values of the time series. The method is useful for time series data that exhibits seasonality and trend patterns. It can provide accurate forecasts even when the data is noisy or has missing values.
TD_HOLT_WINTERS_FORECASTER is used in demand forecasting, financial forecasting, and inventory management.
For single exponential smoothing, see TD_SIMPLEEXP.