EngineersofAI LogoCoursesBlogResearch LabAI LettersThe LabCode BankInteractive 3DKodrEarnestJobs
Skip to main content
Interactive 3D/Stationarity
Series Type
Stationary when |φ| < 1
AR(1) Coefficient φ
φ0.50
Stationary (|φ| < 1)
Parameters
T (length)150
Noise σ1.0
Rolling window30
Statistics
Sample mean: 0.000
Sample variance: 0.000
ADF statistic: 0.000
(ADF < -2.86 → reject unit root)
Stationary: Yes
Key Insight
A stationary series has constant mean and variance over time. Rolling stats stay flat. Non-stationary series have drifting mean or growing variance - standard models break.

Stationarity - Interactive Visualization

Stationarity is the assumption that statistical properties (mean, variance, autocovariance) are constant over time. Most time series models require stationary input. An AR(1) process with |φ| < 1 is stationary. A random walk (φ = 1) is non-stationary - it drifts and its variance grows without bound. Rolling mean and variance reveal non-stationarity visually.

  • Compare AR(1) (stationary) vs random walk vs trend series
  • Rolling mean and std overlay reveals drifting properties
  • Adjust AR coefficient φ: approach 1 to see near-unit-root behavior
  • See ADF test statistic change with stationarity
  • Foundation for ARIMA, Granger causality, and financial time series

Part of the EngineersOfAI Interactive 3D - free interactive visualizations covering every major concept in machine learning and AI engineering. Hover any element for a plain-English explanation. No code required.