Model Atlas · Time series

ARIMA

Models autoregressive structure, differencing and moving-average residuals.

core mathematical viewARIMA(p,d,q)
Mental model

Understand it before memorizing it.

Explain today from past values and past forecast errors after stabilizing the series.

Best fit

Where this model earns its place

Classical forecasting

Start with a simpler baseline then compare this model using the same evaluation split and operational constraints.

Stationary-ish signals

Start with a simpler baseline then compare this model using the same evaluation split and operational constraints.

Transparent baselines

Start with a simpler baseline then compare this model using the same evaluation split and operational constraints.

Strengths and limits

Trade-offs matter more than popularity.

Strengths

✓ Interpretable

✓ Good baseline

Limitations

△ Weak with complex nonlinear seasonality

Evaluation

Metrics to watch

MAEInterpret with the product objective and error cost.
RMSEInterpret with the product objective and error cost.
MAPEInterpret with the product objective and error cost.
Production checklist

Before it reaches real users

  1. 01

    Check stationarity

    Document the assumption and instrument the condition so regressions can be detected.

  2. 02

    Backtest chronologically

    Document the assumption and instrument the condition so regressions can be detected.

  3. 03

    Monitor residuals

    Document the assumption and instrument the condition so regressions can be detected.