---

name: datasets-results-graphics description: "Use statsmodels datasets, result objects, predictions, summaries, robust covariance outputs, saving/loading, graphics, I/O, and local-vs-network support workflows." disable-model-invocation: true metadata: disco-role: operating license: BSD 3-Clause

Datasets, results, and graphics

Use this sub-skill for support workflows around statsmodels package use: built-in datasets, get_rdataset network behavior, result object attributes, prediction frames, summary/export, saving/loading fitted results, diagnostic/statistical graphics, table I/O, and webdoc.

Workflow

1. Prefer built-in datasets or user-provided local data for reproducible examples. Treat get_rdataset as network-dependent.
2. Fit the model through the owning model sub-skill, then use result attributes for computation and summaries/plots for presentation.
3. For graphics in automation, set a noninteractive Matplotlib backend before importing pyplot.
4. Use bundled smoke helpers for minimal environment/result checks instead of broad example runners.

Read or run

• Read references/data-and-results.md for built-in datasets, result attributes, predictions, summaries, robust covariance, and save/load.
• Read references/graphics-and-io.md for plot functions, headless plotting, summary export, pickle/I/O, and webdoc.
• Read references/troubleshooting.md for network datasets, plot backend, prediction shape, summary-as-data, and pickle compatibility pitfalls.
• Run scripts/smoke_results_graphics.py for a local Longley/synthetic fit, prediction, summary, and headless figure smoke check.

Boundaries

• Route model choice and fit setup to linear-and-formula-models, discrete-and-count-models, or time-series-analysis.
• Route hypothesis-test choice and residual diagnostic interpretation to statistical-tests-and-diagnostics.
• This sub-skill owns how to operationalize outputs after fitting: tables, predictions, plots, serialization, and data source safety.