STOP 13 / 31Let recent observations speak louder.
Layout
STOP 13 · Models

Let recent observations speak louder.

EXPLORE~6 min
Back to route map
  • Simple exponential smoothing (SES) updates a level estimate each time a new observation arrives.
  • Recent data receive more weight, with older observations fading exponentially.
Simple exponential smoothing: concepts and Northstar examples. Illustrative Northstar examples, not measured business results.
ConceptWhat it meansNorthstar example
Alpha controls the reaction speed.
  • A high α quickly follows new observations—and noise.
  • A low α is steadier but slower to adapt.
  • Choose it using training data and appropriate evaluation.

A high alpha reacts quickly if Northstar's ordinary sales level changes, but it can also overreact to one unusually busy day.

The forecast is flat.
  • SES is for a series without clear trend or seasonality.
  • Its multi-step point forecasts equal the latest estimated level.

On a suitably adjusted series with no clear trend, SES projects Northstar's latest estimated level forward rather than inventing a seasonal rhythm.

Don’t smooth away a known event.
  • A Black Friday peak should not permanently reset Northstar’s ordinary-day level.
  • Model recurring structure instead of forcing SES to absorb it.

Treat Black Friday as a known event instead of allowing its sales peak to permanently reset Northstar's ordinary-day level.

levelₜ = αyₜ + (1 − α)levelₜ₋₁
FIG 13.1

How quickly should a model react?

WHAT THIS SHOWS

Move alpha to change the weight given to Northstar's latest observation. The level update is computed recursively; additional supplied trend illustrations are labeled separately.

The question it answers: How does a faster reaction trade off against following noise?

Interactive time series chart$23k$27k$31k$35kTodayHistoryLevel shift+20 days
Observed salesSmoothed levelIllustrative forecast

Higher α follows the level shift faster but also follows noise. This playground computes SES level updates, not a full seasonal ETS fit. SES point forecasts are flat beyond the last observation.

Fine print & assumptions

Constructed teaching data and illustrative calculations, not a production forecasting service. Read the diagram's method notes and distinguish supplied simulator components from fitted estimates.

CHECK YOURSELF · CHOOSE AN ANSWER

What happens when α increases?

Your learning note

Private · just for you