Simple Exponential Smoothing (SES)

Simple Exponential Smoothing (SES) is a forecasting technique that only tracks the level of the data. It is designed for time series data that does not contain trend or seasonality.

SES requires only one smoothing parameter:

• Alpha (α) → Level smoothing factor

It does not require: - Frequency - Seasonal period

Key Characteristics

Level-Only Model

SES assumes that the data represents a constant underlying level, meaning:

• No upward or downward trend
• No repeating seasonal pattern
• Forecast is based only on smoothed average values

No Trend or Seasonality

• ❌ Beta (β) → Trend component is not used
• ❌ Gamma (γ) → Seasonality component is not used

If these parameters are provided in functions like ses(), they are ignored, because SES assumes the data is purely level-based.

Forecast Behavior

Flat Forecast Line

When SES is applied:

• The forecast output (e.g., wframe1) is a perfectly horizontal line
• Future values are assumed to remain constant
• Best prediction = most recently smoothed level

Effect of Alpha (α)

The smoothing parameter Alpha (α) controls how quickly the model reacts to changes in the data.

High Alpha (e.g., 0.9)

• Gives more weight to recent observations
• Forecast reacts quickly to changes
• The horizontal forecast line starts close to the last observed data point

Low Alpha (e.g., 0.1)

• Gives more weight to historical data
• Produces smoother forecasts
• The forecast line stays closer to the long-term average of the dataset