Simple Exponential Smoothing (SES)
You'll learn how SES uses a weighted average to forecast time series data that lacks trend or seasonality.
What is Simple Exponential Smoothing?
Simple Exponential Smoothing (SES) is a forecasting method that creates predictions by taking a weighted average of past observations, where recent observations get more weight than older ones. Unlike a simple moving average that treats all observations in the window equally, SES gives exponentially decreasing importance to older data points.
Think of it like your memory: you naturally remember yesterday's events more vividly than what happened a month ago. SES formalizes this intuition mathematically.
The SES Formula
The forecast at time t+1 is calculated as:
Forecast(t+1) = α × Actual(t) + (1 - α) × Forecast(t)
Where:
- α (alpha) is the smoothing parameter between 0 and 1
- Actual(t) is the most recent observed value
- Forecast(t) is the previous forecast
Understanding the Smoothing Parameter α
The value of α controls how much weight goes to recent vs historical data:
- α close to 1 (e.g., 0.8): Heavy weight on recent observations → forecast responds quickly to changes
- α close to 0 (e.g., 0.2): More weight on historical average → forecast is smoother and more stable
When to Use SES
Use Simple Exponential Smoothing when your time series has:
- ✅ No trend (no consistent upward or downward movement)
- ✅ No seasonality (no repeating patterns)
- ✅ Fairly stable level with random fluctuations
Key Takeaway: SES is ideal for stationary data without trend or seasonality, using the smoothing parameter α to balance responsiveness to recent changes against historical stability.