---

name: clustering description: "Provides elbow-curve analysis for clusterers and direct cluster-count workflows in scikit-plot." disable-model-invocation: true metadata: disco-role: operating license: MIT

Clustering

Use this sub-skill when the user wants to choose K, compare clustering inertia across cluster counts, or render an elbow curve for a cloneable scikit-learn-style clusterer.

What this covers

• scikitplot.cluster.plot_elbow_curve
• elbow-curve analysis for KMeans-like estimators
• cluster-count sweeps over a finite cluster_ranges
• optional elapsed-time overlays and Matplotlib Axes reuse

Route elsewhere when

• you need deprecated instance-method injection such as clustering_factory -> legacy-factories
• you need silhouette plots from fitted labels or label-based clustering metrics -> metrics
• you need feature-importance or learning-curve routes -> estimators
• you need decomposition or PCA routes -> decomposition

Use this route

1. Start from scikitplot.cluster.plot_elbow_curve.
2. Confirm the clusterer is clone-compatible and exposes n_clusters, fit, and score.
3. Pick a small finite cluster range such as range(1, 11).
4. Keep n_jobs=1 for the most predictable smoke run unless you are explicitly checking the parallel path.
5. Set show_cluster_time=True only when you want the timing overlay.
6. Pass ax= only when embedding into an existing figure.
7. Use ax.figure to save or close the result if needed.

Read next

• references/api-reference.md for the exact function contract and implementation notes.
• references/workflows.md for K-selection and comparison recipes.
• references/troubleshooting.md for interface, range, parallelism, and import failures.
• scripts/clustering_smoke.py for a tiny Agg-backed elbow-curve smoke run.