← Lessons

quiz vs the machine

Silver1040

Machine Learning

Time Series Components Trend And Seasonality

Breaking a series into trend, seasonal, and residual parts to understand its shape.

4 min read · intro · beat Silver to climb

Time Series Components Trend And Seasonality

A time series is a sequence of values recorded over time. To reason about it, we decompose the signal into a few interpretable parts.

The three parts

  • Trend: the slow long term direction, such as a steady rise in sales over years.
  • Seasonality: a repeating pattern with a fixed period, such as higher ice cream demand every summer.
  • Residual: the leftover noise after trend and seasonality are removed.

Additive and multiplicative

Two common ways to combine the parts:

  • Additive: value equals trend plus seasonality plus residual. Use this when the seasonal swing stays roughly constant in size.
  • Multiplicative: value equals trend times seasonality times residual. Use this when the swing grows with the level of the series.

Why decompose

Separating the parts lets you describe the trend cleanly, model the season, and study the residual for surprises. Many forecasting methods either remove these components first or model each one directly.

Key idea

Decomposing a series into trend, seasonality, and residual turns a tangled line into parts you can model and explain.

Check yourself

Answer to earn rating on the learn ladder.

1. What is seasonality in a time series?

2. When is a multiplicative model preferred?