TES (Triple Exponential Smoothing)
TES (Triple Exponential Smoothing) is the most complete model in the exponential smoothing family. It captures Level, Trend, and Seasonality simultaneously, making it suitable for real-world time series forecasting.
TES is used when a dataset contains: - A baseline level (Level) - A directional movement (Trend) - A repeating pattern (Seasonality)
It is commonly implemented using Holt-Winters / ETS models.
TES models a time series using three components:
Forecast = Level + Trend + Seasonality (Additive form)
OR
Forecast = (Level + Trend) × Seasonality (Multiplicative form)