DESLT (Double Exponential Smoothing with Linear Trend)

DESLT (Double Exponential Smoothing with Linear Trend) is a time series forecasting technique used when data shows a trend but no seasonality.

It extends Simple Exponential Smoothing by adding a trend component, allowing forecasts to follow a linear pattern.

Key Characteristics

• Uses two smoothing parameters:
• α (alpha) → level smoothing

• β (beta) → trend smoothing

• Produces linear (straight-line) forecasts

• Best suited for non-seasonal trend data
• Lightweight and computationally efficient

Parameters Behavior

α (Alpha) — Level

• Controls how much weight is given to recent observations
• High α (0.9) → very reactive, follows recent data closely
• Low α (0.1) → very smooth, slow to react

β (Beta) — Trend

• Controls how quickly trend changes are adapted
• High β (0.8) → fast-changing trend
• Low β (0.1) → stable trend

DESLT is equivalent to Holt’s Linear Trend Method, designed for data with trend but no seasonality.

Unlike ARIMA, which requires parameters (p, d, q), DESLT relies on smoothing parameters (α and β), which are automatically optimized in holt().

Although ets(data, model="AAN") also performs the same method (Additive Error, Additive Trend, No Seasonality), holt() is a direct and simplified wrapper specifically designed for DESLT.