Double Exponential Smoothing (DES)

Double Exponential Smoothing (DES), also known as Holt’s Linear Trend Method, is used for forecasting time series data with trend but no seasonality.

It uses two smoothing parameters: - Alpha (α): controls level smoothing
- Beta (β): controls trend smoothing

Key Points

• Captures trend (upward or downward) in data
• Automatically adapts trend direction using β
• Produces a linear (straight-line) forecast
• Best suited for short to medium-term forecasting

Limitation

• Assumes a constant linear trend
• Can lead to overshooting in long-term forecasts
• May produce unrealistic continuous growth or decline

Improvement

To overcome this limitation, use Damped Trend Holt’s Method, which introduces a phi (φ) parameter to gradually reduce the trend effect and flatten forecasts over time.