Plotting¶

Requires the plot extra: pip install "scikit-opls[plot]".

Diagnostic plots for OPLS models.

The public surface follows scikit-learn's Display convention: each plot is a class with a :meth:from_estimator constructor that computes the plotted arrays, a :meth:plot method that draws them, and stored ax_ / figure_ handles.

matplotlib is an optional dependency (pip install scikit-opls[plot]) and is imported lazily inside :meth:plot, so importing this module never requires it.

OPLSScoresDisplay ¶

OPLSScoresDisplay(
    *,
    t_predictive: NDArray[float64],
    t_orthogonal: NDArray[float64],
    y: NDArray | None = None,
    predictive_component: int = 0,
    orthogonal_component: int = 0,
    has_orthogonal: bool = True,
)

Predictive vs orthogonal score scatter for an OPLS-family model.

Works for :class:~scikit_opls.OPLS, :class:~scikit_opls.OPLSDA, a pipeline ending in one, or a fitted search meta-estimator exposing best_estimator_ around either shape. Construct with :meth:from_estimator.

Parameters:

Name	Type	Description	Default
`t_predictive`	`ndarray of shape (n_samples,)`	Selected predictive score per sample.	required
`t_orthogonal`	`ndarray of shape (n_samples,)`	Selected orthogonal score per sample (zeros when the model has none).	required
`y`	`ndarray of shape (n_samples,) or None`	Optional labels used to colour the points.	`None`

Attributes:

Name	Type	Description
`ax_`	`matplotlib Axes`	The axes drawn on (set by :meth:`plot`).
`figure_`	`matplotlib Figure`	The parent figure (set by :meth:`plot`).
`scatter_`	`matplotlib PathCollection or list of PathCollection`	The scatter plot artist(s) (set by :meth:`plot`).
`has_orthogonal`	`bool`	Whether the fitted model has any orthogonal component. When `False` the orthogonal axis is a constant-zero placeholder and labelled as such.

from_estimator `classmethod` ¶

from_estimator(
    estimator: BaseEstimator,
    X: ArrayLike,
    y: ArrayLike | None = None,
    *,
    predictive_component: int = 0,
    orthogonal_component: int = 0,
    ax: Axes | None = None,
) -> OPLSScoresDisplay

Compute the scores from a fitted estimator and plot them.

Parameters:

Name	Type	Description	Default
`estimator`	`OPLS, OPLSDA, Pipeline or search meta-estimator`	A fitted estimator. When passing a pipeline, pass raw `X` as expected by that pipeline. When passing the final OPLS-family step directly, pass the already transformed matrix seen by that step.	required
`X`	`array-like of shape (n_samples, n_features)`	Samples to project.	required
`y`	`array-like of shape (n_samples,)`	Optional labels used to colour the points.	`None`
`predictive_component`	`int`	The index of the predictive PLS component to plot.	`0`
`orthogonal_component`	`int`	The index of the orthogonal component to plot.	`0`
`ax`	`matplotlib Axes`	Target axes; a new figure/axes is created when `None`.	`None`

Returns:

Name	Type	Description
`display`	`OPLSScoresDisplay`	The plotted display, with `ax_` / `figure_` set.

plot ¶

plot(ax: Axes | None = None) -> OPLSScoresDisplay

Draw the score scatter on ax (or a fresh axes).

Parameters:

Name	Type	Description	Default
`ax`	`matplotlib Axes`	Target axes; a new figure/axes is created when `None`.	`None`

Returns:

Name	Type	Description
`display`	`OPLSScoresDisplay`	`self`, with `ax_` / `figure_` populated.

SPlotDisplay ¶

SPlotDisplay(
    *,
    covariance: NDArray[float64],
    correlation: NDArray[float64],
    component: int = 0,
)

S-plot: covariance vs correlation of each feature with the predictive score.

.. note:: SPlotDisplay is intended for the fitted training data. For x_space="centered", X is centered by the fitted model mean. For x_space="scaled", X is centered/scaled by the fitted model preprocessing. For x_space="subset-centered", X is centered by the provided subset mean.

When ``estimator`` is a pipeline, the S-plot is computed in the feature
space received by the final OPLS-family step after upstream transformations.
Points may therefore represent transformed features rather than original
input columns.

Accepts :class:~scikit_opls.OPLS, :class:~scikit_opls.OPLSDA, a pipeline ending in one, or a fitted search meta-estimator exposing best_estimator_ around either shape. Construct with :meth:from_estimator.

Parameters:

Name	Type	Description	Default
`covariance`	`ndarray of shape (n_features,)`	Covariance of each feature with the selected predictive score.	required
`correlation`	`ndarray of shape (n_features,)`	Correlation of each feature with the selected predictive score.	required

Attributes:

Name	Type	Description
`ax_`	`matplotlib Axes`	The axes drawn on (set by :meth:`plot`).
`figure_`	`matplotlib Figure`	The parent figure (set by :meth:`plot`).
`scatter_`	`matplotlib PathCollection`	The scatter plot artist (set by :meth:`plot`).

from_estimator `classmethod` ¶

from_estimator(
    estimator: BaseEstimator,
    X: ArrayLike,
    *,
    component: int = 0,
    x_space: str = "centered",
    ax: Axes | None = None,
) -> SPlotDisplay

Compute the S-plot arrays from a fitted estimator and plot them.

Parameters:

Name	Type	Description	Default
`estimator`	`OPLS, OPLSDA, Pipeline or search meta-estimator`	A fitted estimator. When passing a pipeline, pass raw `X` as expected by that pipeline. When passing the final OPLS-family step directly, pass the already transformed matrix seen by that step.	required
`X`	`array-like of shape (n_samples, n_features)`	Samples to project.	required
`component`	`int`	The index of the predictive PLS component to plot.	`0`
`x_space`	`('centered', 'scaled', 'subset-centered')`	Feature space used for the covariance/correlation axes. `"centered"` uses original feature units centered by the fitted training mean, `"scaled"` uses the model-scaled feature space, and `"subset-centered"` centers the provided X subset by its own mean.	`"centered"`
`ax`	`matplotlib Axes`	Target axes; a new figure/axes is created when `None`.	`None`

Returns:

Name	Type	Description
`display`	`SPlotDisplay`	The plotted display, with `ax_` / `figure_` set.

plot ¶

plot(ax: Axes | None = None) -> SPlotDisplay

Draw the S-plot scatter on ax (or a fresh axes).

Parameters:

Name	Type	Description	Default
`ax`	`matplotlib Axes`	Target axes; a new figure/axes is created when `None`.	`None`

Returns:

Name	Type	Description
`display`	`SPlotDisplay`	`self`, with `ax_` / `figure_` populated.

options: members:

OPLSScoresDisplay
SPlotDisplay

Plotting¶

OPLSScoresDisplay ¶

from_estimator classmethod ¶

plot ¶

SPlotDisplay ¶

from_estimator classmethod ¶

plot ¶

from_estimator `classmethod` ¶

from_estimator `classmethod` ¶