Course contentsShow
← Data Science
Lesson 712 of 2,14519. Time Series FoundationsFree lesson

What is Stationarity?

Define stationarity as constant statistical properties over time, including constant mean, variance, and autocorrelation structure.

What is Stationarity?

You'll learn: What it means for a time series to be stationary and why this property matters for analysis.

The Core Idea

Stationarity means that a time series has constant statistical properties over time. Specifically, a stationary time series maintains:

  1. Constant mean — the average value doesn't drift up or down
  2. Constant variance — the spread or volatility stays the same
  3. Constant autocorrelation structure — the relationship between observations at different time lags remains stable

Think of stationarity like a calm ocean with consistent wave patterns. The waves might go up and down, but the average water level stays the same, the size of typical waves doesn't change, and the rhythm between waves remains steady. A non-stationary series is more like tides coming in and out — the water level itself changes over time.

Why Does This Matter?

Many time series models and statistical techniques assume stationarity. When a series has a trend (like rising sales over years), seasonal patterns that grow stronger, or changing volatility, it violates these assumptions. You've already learned about trend, seasonality, and other components — stationarity is about whether those components shift the fundamental statistical properties over time.

For example:

  • A time series with no trend and constant seasonality can be stationary
  • A time series with an upward trend is not stationary (mean changes)
  • Stock prices with increasing volatility are not stationary (variance changes)

Key Takeaway: Stationarity means constant mean, variance, and autocorrelation over time — a stable statistical "baseline" that makes time series analysis more reliable.