plot_habitat_feature_components

Note

This page is a reference documentation. It only explains the function signature, and not how to use it. Please refer to the Habitat Guide and Python API guide (v2.0) for usage.

plot_habitat_feature_components(data: 'HabitatFeaturePanel' | 'HabitatFeatureComparison', *, method: Literal['pca', 'cva'] = 'cva', n_components: int = 2, features: Sequence[str] | None = None, habitats: Sequence[int] | None = None, annotate_subjects: bool | None = None, show_loadings: bool = True, title: str | None = None) → Figure[source]

Habitat contrast on a few CVA or PCA component scores.

This is the overflow figure when the pair x feature Cliff’s-delta heatmap would be too tall: compute a few axes, then show how H1..Hk differ on those scores (same job as the delta / bar plots; the “features” are CV1/PC1/…). It is not a 2-D embedding to admire.

Default method="cva" is multi-class Fisher LDA (canonical variates that separate habitats) – not two-block CCA. When n_features >= n_samples - n_classes the CVA path reduces with PCA first and the title says CVA (PCA-preprocessed). Pass method="pca" for unsupervised components.

Features are z-scored before the fit so Energy and volume_fraction share one Euclidean space. A loadings row names the original features that represent each retained axis.

Parameters:

  • data – Long panel or a HabitatFeatureComparison.

  • method – "cva" (default) or "pca".

  • n_components – Requested axes. CVA keeps at most n_habitats - 1.

  • features – Optional feature subset. Default: all panel features.

  • habitats – Optional habitat subset.

  • annotate_subjects – Unused. Kept so existing callers that passed it still construct; the figure is habitat contrast on scores, not a subject scatter.

  • show_loadings – Draw one loadings panel per retained component.

  • title – Optional figure title. Default names the contrast and the p features -> k axes reduction.

Returns:

The matplotlib Figure.

Raises:

HABITAPIError – Unknown method, empty panel, or a still- singular CVA after PCA.