Mixins

Mixins to support the plotting and slicing of the distribution parameters. They are only required for creating new distributions or for understanding the internals.

`ContinuousPlotDistMixin`

Bases: PlotDistMixin

Functionality for plot_pdf method of continuous distributions.

Source code in conjugate/plot.py

class ContinuousPlotDistMixin(PlotDistMixin):
    """Functionality for plot_pdf method of continuous distributions."""

    def _plot(self, ax: Axes | None = None, cdf: bool = False, **kwargs) -> Axes:
        x = self._create_x_values()
        x = self._reshape_x_values(x)

        ax = self._settle_axis(ax=ax)

        return self._create_plot_on_axis(x=x, cdf=cdf, ax=ax, **kwargs)

    def plot_pdf(self, ax: Axes | None = None, **kwargs) -> Axes:
        """Plot the PDF of distribution

        Args:
            ax: matplotlib Axes, optional
            **kwargs: Additonal kwargs to pass to matplotlib

        Returns:
            new or modified Axes

        Raises:
            ValueError: If the max_value is not set.

        """
        return self._plot(ax=ax, cdf=False, **kwargs)

    def plot_cdf(self, ax: Axes | None = None, **kwargs) -> Axes:
        """Plot the CDF of distribution

        Args:
            ax: matplotlib Axes, optional
            **kwargs: Additonal kwargs to pass to matplotlib

        Returns:
            new or modified Axes

        Raises:
            ValueError: If the max_value is not set.

        """
        return self._plot(ax=ax, cdf=True, **kwargs)

    def _create_x_values(self) -> np.ndarray:
        return np.linspace(self.min_value, self.max_value, 100)

    def _setup_labels(self, ax, cdf: bool = False) -> None:
        if isinstance(ax, PolarAxes):
            return

        ylabel = "Density $f(x)$" if not cdf else "Cumulative Density $F(x)$"
        ax.set(xlabel="Domain", ylabel=ylabel)

    def _create_plot_on_axis(self, x, cdf: bool, ax: Axes, **kwargs) -> Axes:
        func = self.dist.cdf if cdf else self.dist.pdf
        yy = func(x)

        if "label" in kwargs:
            label = kwargs.pop("label")
            label = resolve_label(label, yy)
        else:
            label = None

        if "color" in kwargs and isinstance(kwargs["color"], Iterable):
            ax.set_prop_cycle(color=kwargs.pop("color"))

        ax.plot(x, yy, label=label, **kwargs)
        self._setup_labels(ax=ax, cdf=cdf)
        ax.set_ylim(0, None)
        return ax

`plot_cdf(ax=None, **kwargs)`

Plot the CDF of distribution

Parameters:

Name	Type	Description	Default
`ax`	`Axes \| None`	matplotlib Axes, optional	`None`
`**kwargs`		Additonal kwargs to pass to matplotlib	`{}`

Returns:

Type	Description
`Axes`	new or modified Axes

Raises:

Type	Description
`ValueError`	If the max_value is not set.

Source code in conjugate/plot.py

def plot_cdf(self, ax: Axes | None = None, **kwargs) -> Axes:
    """Plot the CDF of distribution

    Args:
        ax: matplotlib Axes, optional
        **kwargs: Additonal kwargs to pass to matplotlib

    Returns:
        new or modified Axes

    Raises:
        ValueError: If the max_value is not set.

    """
    return self._plot(ax=ax, cdf=True, **kwargs)

`plot_pdf(ax=None, **kwargs)`

Plot the PDF of distribution

Parameters:

Name	Type	Description	Default
`ax`	`Axes \| None`	matplotlib Axes, optional	`None`
`**kwargs`		Additonal kwargs to pass to matplotlib	`{}`

Returns:

Type	Description
`Axes`	new or modified Axes

Raises:

Type	Description
`ValueError`	If the max_value is not set.

Source code in conjugate/plot.py

def plot_pdf(self, ax: Axes | None = None, **kwargs) -> Axes:
    """Plot the PDF of distribution

    Args:
        ax: matplotlib Axes, optional
        **kwargs: Additonal kwargs to pass to matplotlib

    Returns:
        new or modified Axes

    Raises:
        ValueError: If the max_value is not set.

    """
    return self._plot(ax=ax, cdf=False, **kwargs)

`DirichletPlotDistMixin`

Bases: ContinuousPlotDistMixin

Plot the pdf using samples from the dirichlet distribution.

Source code in conjugate/plot.py

class DirichletPlotDistMixin(ContinuousPlotDistMixin):
    """Plot the pdf using samples from the dirichlet distribution."""

    def plot_pdf(
        self,
        ax: Axes | None = None,
        samples: int = 1_000,
        random_state=None,
        **kwargs,
    ) -> Axes:
        """Plots the pdf by sampling from the distribution.

        Args:
            ax: matplotlib Axes, optional
            samples: number of samples to take from the distribution
            random_state: random state to use for sampling
            **kwargs: Additonal kwargs to pass to matplotlib

        Returns:
            new or modified Axes

        """
        distribution_samples = self.dist.rvs(size=samples, random_state=random_state)

        ax = self._settle_axis(ax=ax)
        xx = self._create_x_values()

        labels = label_to_iterable(
            kwargs.pop("label", None), distribution_samples.shape[1]
        )

        for x, label in zip_longest(distribution_samples.T, labels):
            kde = gaussian_kde(x)

            yy = kde(xx)
            ax.plot(xx, yy, label=label, **kwargs)

        self._setup_labels(ax=ax)
        return ax

`plot_pdf(ax=None, samples=1000, random_state=None, **kwargs)`

Plots the pdf by sampling from the distribution.

Parameters:

Name	Type	Description	Default
`ax`	`Axes \| None`	matplotlib Axes, optional	`None`
`samples`	`int`	number of samples to take from the distribution	`1000`
`random_state`		random state to use for sampling	`None`
`**kwargs`		Additonal kwargs to pass to matplotlib	`{}`

Returns:

Type	Description
`Axes`	new or modified Axes

Source code in conjugate/plot.py

def plot_pdf(
    self,
    ax: Axes | None = None,
    samples: int = 1_000,
    random_state=None,
    **kwargs,
) -> Axes:
    """Plots the pdf by sampling from the distribution.

    Args:
        ax: matplotlib Axes, optional
        samples: number of samples to take from the distribution
        random_state: random state to use for sampling
        **kwargs: Additonal kwargs to pass to matplotlib

    Returns:
        new or modified Axes

    """
    distribution_samples = self.dist.rvs(size=samples, random_state=random_state)

    ax = self._settle_axis(ax=ax)
    xx = self._create_x_values()

    labels = label_to_iterable(
        kwargs.pop("label", None), distribution_samples.shape[1]
    )

    for x, label in zip_longest(distribution_samples.T, labels):
        kde = gaussian_kde(x)

        yy = kde(xx)
        ax.plot(xx, yy, label=label, **kwargs)

    self._setup_labels(ax=ax)
    return ax

`DiscretePlotMixin`

Bases: PlotDistMixin

Adding the plot_pmf method to class.

Source code in conjugate/plot.py

class DiscretePlotMixin(PlotDistMixin):
    """Adding the plot_pmf method to class."""

    def _plot(
        self,
        ax: Axes | None = None,
        cdf: bool = False,
        mark: str = "o-",
        conditional: bool = False,
        **kwargs,
    ) -> Axes:
        x = self._create_x_values()
        x = self._reshape_x_values(x)

        ax = self._settle_axis(ax=ax)
        return self._create_plot_on_axis(
            x,
            ax=ax,
            cdf=cdf,
            mark=mark,
            conditional=conditional,
            **kwargs,
        )

    def plot_pmf(
        self,
        ax: Axes | None = None,
        mark: str = "o-",
        conditional: bool = False,
        **kwargs,
    ) -> Axes:
        """Plot the PMF of distribution

        Args:
            ax: matplotlib Axes, optional
            mark: matplotlib line style
            conditional: If True, plot the conditional probability given the bounds.
            **kwargs: Additonal kwargs to pass to matplotlib

        Returns:
            new or modified Axes

        Raises:
            ValueError: If the max_value is not set.

        """
        return self._plot(
            ax=ax,
            cdf=False,
            mark=mark,
            conditional=conditional,
            **kwargs,
        )

    def plot_cdf(
        self,
        ax: Axes | None = None,
        mark: str = "o-",
        conditional: bool = False,
        **kwargs,
    ) -> Axes:
        """Plot the CDF of distribution

        Args:
            ax: matplotlib Axes, optional
            mark: matplotlib line style
            conditional: If True, plot the conditional probability given the bounds.
            **kwargs: Additonal kwargs to pass to matplotlib

        Returns:
            new or modified Axes

        Raises:
            ValueError: If the max_value is not set.

        """
        return self._plot(ax=ax, cdf=True, mark=mark, conditional=conditional, **kwargs)

    def _create_x_values(self) -> np.ndarray:
        return np.arange(self.min_value, self.max_value + 1, 1)

    def _create_plot_on_axis(
        self,
        x,
        ax,
        cdf: bool,
        mark,
        conditional: bool = False,
        **kwargs,
    ) -> Axes:
        func = self.dist.cdf if cdf else self.dist.pmf
        yy = func(x)

        if conditional:
            yy = yy / np.sum(yy)

            prefix = (
                "Cumulative Probability $F(X \\leq x" if cdf else "Probability $f(x"
            )

            ylabel = f"Conditional {prefix}|{self.min_value} \\leq x \\leq {self.max_value})$"
        else:
            ylabel = (
                "Cumulative Probability $F(X \\leq x)$" if cdf else "Probability $f(x)$"
            )

        if "label" in kwargs:
            label = kwargs.pop("label")
            label = resolve_label(label, yy)
        else:
            label = None

        if "color" in kwargs and isinstance(kwargs["color"], Iterable):
            ax.set_prop_cycle(color=kwargs.pop("color"))

        ax.plot(x, yy, mark, label=label, **kwargs)

        if self.max_value - self.min_value < 15:
            ax.set_xticks(x.ravel())
        else:
            ax.set_xticks(x.ravel(), minor=True)
            ax.set_xticks(x[::5].ravel())

        ax.set_xlabel("Domain")
        ax.set_ylabel(ylabel)
        ax.set_ylim(0, None)
        return ax

`plot_cdf(ax=None, mark='o-', conditional=False, **kwargs)`

Plot the CDF of distribution

Parameters:

Name	Type	Description	Default
`ax`	`Axes \| None`	matplotlib Axes, optional	`None`
`mark`	`str`	matplotlib line style	`'o-'`
`conditional`	`bool`	If True, plot the conditional probability given the bounds.	`False`
`**kwargs`		Additonal kwargs to pass to matplotlib	`{}`

Returns:

Type	Description
`Axes`	new or modified Axes

Raises:

Type	Description
`ValueError`	If the max_value is not set.

Source code in conjugate/plot.py

def plot_cdf(
    self,
    ax: Axes | None = None,
    mark: str = "o-",
    conditional: bool = False,
    **kwargs,
) -> Axes:
    """Plot the CDF of distribution

    Args:
        ax: matplotlib Axes, optional
        mark: matplotlib line style
        conditional: If True, plot the conditional probability given the bounds.
        **kwargs: Additonal kwargs to pass to matplotlib

    Returns:
        new or modified Axes

    Raises:
        ValueError: If the max_value is not set.

    """
    return self._plot(ax=ax, cdf=True, mark=mark, conditional=conditional, **kwargs)

`plot_pmf(ax=None, mark='o-', conditional=False, **kwargs)`

Plot the PMF of distribution

Parameters:

Name	Type	Description	Default
`ax`	`Axes \| None`	matplotlib Axes, optional	`None`
`mark`	`str`	matplotlib line style	`'o-'`
`conditional`	`bool`	If True, plot the conditional probability given the bounds.	`False`
`**kwargs`		Additonal kwargs to pass to matplotlib	`{}`

Returns:

Type	Description
`Axes`	new or modified Axes

Raises:

Type	Description
`ValueError`	If the max_value is not set.

Source code in conjugate/plot.py

def plot_pmf(
    self,
    ax: Axes | None = None,
    mark: str = "o-",
    conditional: bool = False,
    **kwargs,
) -> Axes:
    """Plot the PMF of distribution

    Args:
        ax: matplotlib Axes, optional
        mark: matplotlib line style
        conditional: If True, plot the conditional probability given the bounds.
        **kwargs: Additonal kwargs to pass to matplotlib

    Returns:
        new or modified Axes

    Raises:
        ValueError: If the max_value is not set.

    """
    return self._plot(
        ax=ax,
        cdf=False,
        mark=mark,
        conditional=conditional,
        **kwargs,
    )

`PlotDistMixin`

Base mixin in order to support plotting. Requires the dist attribute of the scipy distribution.

Source code in conjugate/plot.py

class PlotDistMixin:
    """Base mixin in order to support plotting. Requires the dist attribute of the scipy distribution."""

    @property
    def dist(self) -> Distribution:
        raise NotImplementedError(
            "Implement this property in the subclass.",
        )  # pragma: no cover

    @property
    def max_value(self) -> float:
        if not hasattr(self, "_max_value"):
            raise ValueError("Set the max value before plotting.")

        return self._max_value

    @max_value.setter
    def max_value(self, value: float) -> None:
        self._max_value = value

    def set_max_value(self, value: float) -> "PlotDistMixin":
        self.max_value = value

        return self

    @property
    def min_value(self) -> float:
        if not hasattr(self, "_min_value"):
            self._min_value = 0.0

        return self._min_value

    @min_value.setter
    def min_value(self, value: float) -> None:
        self._min_value = value

    def set_min_value(self, value: float) -> "PlotDistMixin":
        """Set the minimum value for plotting."""
        self.min_value = value

        return self

    def set_bounds(self, lower: float, upper: float) -> "PlotDistMixin":
        """Set both the min and max values for plotting."""
        return self.set_min_value(lower).set_max_value(upper)

    def _reshape_x_values(self, x: np.ndarray) -> np.ndarray:
        """Make sure that the values are ready for plotting."""
        for value in asdict(self).values():
            if not isinstance(value, float):
                return x[:, None]

        return x

    def _settle_axis(self, ax: Axes | None = None) -> Axes:
        return ax if ax is not None else plt.gca()

`set_bounds(lower, upper)`

Set both the min and max values for plotting.

Source code in conjugate/plot.py

def set_bounds(self, lower: float, upper: float) -> "PlotDistMixin":
    """Set both the min and max values for plotting."""
    return self.set_min_value(lower).set_max_value(upper)

`set_min_value(value)`

Set the minimum value for plotting.

Source code in conjugate/plot.py

def set_min_value(self, value: float) -> "PlotDistMixin":
    """Set the minimum value for plotting."""
    self.min_value = value

    return self

`resolve_label(label, yy)`

https://stackoverflow.com/questions/73662931/matplotlib-plot-a-numpy-array-as-many-lines-with-a-single-label

Source code in conjugate/plot.py

def resolve_label(label: LABEL_INPUT, yy: np.ndarray):
    """

    https://stackoverflow.com/questions/73662931/matplotlib-plot-a-numpy-array-as-many-lines-with-a-single-label
    """
    if yy.ndim == 1:
        return label

    ncols = yy.shape[1]
    if ncols != 1:
        return label_to_iterable(label, ncols)

    return label

`SliceMixin`

Mixin in order to slice the parameters

Source code in conjugate/slice.py

class SliceMixin:
    """Mixin in order to slice the parameters"""

    @property
    def params(self):
        return asdict(self)

    def __getitem__(self, key):
        def take_slice(value, key):
            try:
                return value[key]
            except Exception:
                return value

        new_params = {k: take_slice(value=v, key=key) for k, v in self.params.items()}

        return self.__class__(**new_params)

Mixins

ContinuousPlotDistMixin

plot_cdf(ax=None, **kwargs)

plot_pdf(ax=None, **kwargs)

DirichletPlotDistMixin

plot_pdf(ax=None, samples=1000, random_state=None, **kwargs)

DiscretePlotMixin

plot_cdf(ax=None, mark='o-', conditional=False, **kwargs)

plot_pmf(ax=None, mark='o-', conditional=False, **kwargs)

PlotDistMixin

set_bounds(lower, upper)

set_min_value(value)

resolve_label(label, yy)

SliceMixin

Comments

`ContinuousPlotDistMixin`

`plot_cdf(ax=None, **kwargs)`

`plot_pdf(ax=None, **kwargs)`

`DirichletPlotDistMixin`

`plot_pdf(ax=None, samples=1000, random_state=None, **kwargs)`

`DiscretePlotMixin`

`plot_cdf(ax=None, mark='o-', conditional=False, **kwargs)`

`plot_pmf(ax=None, mark='o-', conditional=False, **kwargs)`

`PlotDistMixin`

`set_bounds(lower, upper)`

`set_min_value(value)`

`resolve_label(label, yy)`

`SliceMixin`