Supply Chain LabSeasonality & Trend DecompositionLab Home
Forecasting & Analytics

Seasonality & Trend Decomposition

Pull a demand series apart into trend + seasonal + noise - the building blocks of forecasting

Time Series Decomposition

Observed = Trend × Seasonal × Noise

Signal Strength

Trend slope /period
0
Seasonal peak index
0
Noise level (CV)
0
Season length
12

Series Controls

About this tool

Seasonality & Trend Decomposition

The concept

A time series splits into trend, seasonal pattern and residual noise.

How it works

Moving averages estimate trend; seasonal indices capture the repeating pattern; what remains is noise.

Winning strategy

Forecast trend and season separately, and never fit the noise.

y_t = Trend × Season × Noise (multiplicative, as drawn here; the additive form is Trend + Season + Residual)
Time seriesTrendSeasonal indexResidual
Core idea: Decomposition of time series
Developed by Dr. Shahrooz Shahparvari · RMIT University · Supply Chain Lab