---

name: time-series-analysis description: "Use statsmodels time-series APIs for stationarity tests, AR/ARIMA/SARIMAX, state-space models, VAR/VECM, exponential smoothing, STL/MSTL, filters, forecasting, and time-series troubleshooting." disable-model-invocation: true metadata: disco-role: operating license: BSD 3-Clause

Time-series analysis

Use this sub-skill for statsmodels.tsa tasks: stationarity/autocorrelation tests, AR/AutoReg, ARDL/UECM, ARIMA/SARIMAX, state-space models, VAR/SVAR/VECM, DynamicFactor, ETS/Holt-Winters, STL/MSTL decomposition, filters, forecasting, Markov switching, and X-13/X-12 integration.

Workflow

1. Validate the time index and frequency. Forecasting is more reliable when a pandas Series/DataFrame has a monotonic date/period index with known frequency.
2. Decide the task: test, filter/decompose, univariate forecast, regression with AR errors/exogenous variables, multivariate VAR/VECM, or custom state-space.
3. Fit a small baseline model first. Inspect convergence, stationarity/invertibility restrictions, residual diagnostics, and forecast intervals.
4. Treat X-13/X-12 as optional external-binary functionality. Do not promise it from package import alone.

Read or run

• Read references/api-reference.md for tsa import names and verified constructor signatures.
• Read references/workflows.md for stationarity checks, ARIMA/SARIMAX, STL/ETS, VAR, and forecast recipes.
• Read references/troubleshooting.md for missing frequency, convergence, stationarity/invertibility, exogenous forecast, and X-13 problems.
• Run scripts/smoke_time_series.py for a deterministic ARIMA/STL/ADF smoke check.

Boundaries

• Route ordinary cross-sectional regression to linear-and-formula-models.
• Route residual tests, Ljung-Box interpretation, or generic hypothesis-test selection to statistical-tests-and-diagnostics when the model itself is not time-series-specific.
• Route plots and result export to datasets-results-graphics after choosing the time-series model.