← Decomposition All techniques Next: ARIMA →

23 — Moving Average & Exponential Smoothing

Three classic forecasters. MA averages the last M values equally. SES uses an exponential weighting (recent matters more) via α. Holt adds a trend term (α + γ).

MAE
RMSE
Choosing: bigger M (or smaller α) = smoother forecast but laggier on real shifts. Smaller M (or bigger α) = nimble but jumpy. Holt is the only one of the three that can forecast a sloped future — the others flatten out.