edsteva.viz.plots.estimates_densities.estimates_densities

estimates_densities_plot

estimates_densities_plot(
    fitted_model: BaseModel,
    probe: BaseProbe = None,
    care_site_level: List[str] = None,
    start_date: Union[datetime, str] = None,
    end_date: Union[datetime, str] = None,
    save_path: str = None,
    vertical_bar_charts_config: Dict[str, str] = None,
    horizontal_bar_charts_config: Dict[str, str] = None,
    chart_style: Dict[str, float] = None,
    y_axis_title: str = None,
    **kwargs
)

Displays the density plot with the associated box plot of each estimate and metric computed in the input model. It can help you to set the thresholds.

PARAMETER	DESCRIPTION
probe	Class describing the completeness predictor \(c(t)\) TYPE: BaseProbe DEFAULT: None
fitted_model	Model with estimates of interest. EXAMPLE: StepFunction Model with \((\hat{t_0}, \hat{c_0})\) TYPE: BaseModel
care_site_level	EXAMPLE: ["Hospital"], ["Hôpital", "UF"] or ["UF", "UH"] TYPE: List[str] DEFAULT: None
start_date	EXAMPLE: "2019-05-01" TYPE: datetime DEFAULT: None
end_date	EXAMPLE: "2021-07-01" TYPE: datetime DEFAULT: None
save_path	Folder path where to save the chart in HTML format. TYPE: str DEFAULT: None
vertical_bar_charts_config	Configuration used to construct the vertical bar charts. TYPE: Dict[str, str] DEFAULT: None
horizontal_bar_charts_config	Configuration used to construct the horizontal bar charts. TYPE: Dict[str, str] DEFAULT: None
chart_style	Configuration used to configure the chart style. EXAMPLE: {"labelFontSize": 13, "titleFontSize": 14} TYPE: Dict[str, float] DEFAULT: None
y_axis_title	Label name for the y axis. TYPE: str DEFAULT: None

Source code in edsteva/viz/plots/estimates_densities/estimates_densities.py

def estimates_densities_plot(
    fitted_model: BaseModel,
    probe: BaseProbe = None,
    care_site_level: List[str] = None,
    start_date: Union[datetime, str] = None,
    end_date: Union[datetime, str] = None,
    save_path: str = None,
    vertical_bar_charts_config: Dict[str, str] = None,
    horizontal_bar_charts_config: Dict[str, str] = None,
    chart_style: Dict[str, float] = None,
    y_axis_title: str = None,
    **kwargs,
):
    r"""Displays the density plot with the associated box plot of each estimate and metric computed in the input model. It can help you to set the thresholds.


    Parameters
    ----------
    probe : BaseProbe
        Class describing the completeness predictor $c(t)$
    fitted_model : BaseModel
        Model with estimates of interest.

        **EXAMPLE**: StepFunction Model with $(\hat{t_0}, \hat{c_0})$
    care_site_level : List[str], optional
        **EXAMPLE**: `["Hospital"]`, `["Hôpital", "UF"]` or `["UF", "UH"]`
    start_date : datetime, optional
        **EXAMPLE**: `"2019-05-01"`
    end_date : datetime, optional
        **EXAMPLE**: `"2021-07-01"`
    save_path : str, optional
        Folder path where to save the chart in HTML format.
    vertical_bar_charts_config: Dict[str, str], optional
        Configuration used to construct the vertical bar charts.
    horizontal_bar_charts_config: Dict[str, str], optional
        Configuration used to construct the horizontal bar charts.
    chart_style: Dict[str, float], optional
        Configuration used to configure the chart style.

        **EXAMPLE**: `{"labelFontSize": 13, "titleFontSize": 14}`
    y_axis_title: str, optional,
        Label name for the y axis.
    """
    alt.data_transformers.disable_max_rows()

    estimates = fitted_model.estimates.copy()
    estimates = filter_data(
        data=estimates,
        table_name="estimates",
        care_site_level=care_site_level,
        **kwargs,
    )
    if probe is not None:
        predictor = probe.predictor.copy()
        # Filter data in predictor not in estimates
        predictor = predictor.merge(
            estimates[list(estimates.columns.intersection(set(predictor.columns)))],
            on=list(estimates.columns.intersection(set(predictor.columns))),
        )
        predictor = filter_data(
            data=predictor,
            start_date=start_date,
            end_date=end_date,
        )
        estimates = probe.add_names_columns(estimates)
        probe_config = deepcopy(probe.get_viz_config("estimates_densities_plot"))
        if vertical_bar_charts_config is None:
            vertical_bar_charts_config = probe_config["vertical_bar_charts"]
        if horizontal_bar_charts_config is None:
            horizontal_bar_charts_config = probe_config["horizontal_bar_charts"]
        if chart_style is None:
            chart_style = probe_config["chart_style"]

    quantitative_estimates = []
    time_estimates = []

    base_estimate = alt.Chart(estimates)
    for estimate in fitted_model._coefs + fitted_model._metrics:
        if estimates[estimate].dtype == float or estimates[estimate].dtype == int:
            max_value = estimates[estimate].max()
            min_value = estimates[estimate].min()
            estimates[estimate] = round(estimates[estimate], 3)

            estimate_density = (
                alt.vconcat(
                    (
                        (
                            base_estimate.transform_density(
                                estimate,
                                as_=[estimate, "Density"],
                                extent=[min_value, max_value],
                            )
                            .mark_area()
                            .encode(
                                x=alt.X("{}:Q".format(estimate), title=None),
                                y=alt.Y("Density:Q", title=y_axis_title),
                            )
                        )
                        + base_estimate.mark_rule(color="red").encode(
                            x="median({}):Q".format(estimate),
                            tooltip=alt.Tooltip("median({}):Q".format(estimate)),
                        )
                    ).properties(width=800, height=300),
                    (
                        base_estimate.mark_tick().encode(
                            x=alt.X("{}:Q".format(estimate), axis=None)
                        )
                    ),
                    spacing=0,
                )
                & (
                    base_estimate.mark_boxplot().encode(
                        x="{}:Q".format(estimate),
                    )
                )
            ).resolve_scale(x="shared")
            quantitative_estimates.append(estimate_density)

        else:
            estimates[estimate] = estimates[estimate].astype("datetime64[ns]")
            estimate_density = (
                (
                    base_estimate.transform_timeunit(
                        estimate="yearmonth({})".format(estimate)
                    )
                    .mark_bar(size=10)
                    .encode(
                        x=alt.X(
                            "{}:T".format(estimate),
                            axis=alt.Axis(
                                tickCount="month",
                                format="%Y, %b",
                                labelAngle=-90,
                            ),
                            title=estimate,
                        ),
                        y=alt.Y(
                            "count({}):Q".format(estimate),
                            axis=alt.Axis(tickMinStep=1),
                            title=y_axis_title,
                        ),
                    )
                )
                + base_estimate.mark_rule(color="red").encode(
                    x="median({}):T".format(estimate),
                    tooltip=alt.Tooltip("median({}):T".format(estimate)),
                )
            ).properties(width=800, height=300)
            time_estimates.append(estimate_density)

    estimates_densities = time_estimates + quantitative_estimates

    chart = reduce(
        lambda estimate_density_1, estimate_density_2: estimate_density_1
        & estimate_density_2,
        estimates_densities,
    )
    if probe is not None:
        base = alt.Chart(predictor)

        horizontal_bar_charts, y_variables_selections = generate_horizontal_bar_charts(
            base=base,
            horizontal_bar_charts_config=horizontal_bar_charts_config,
            predictor=predictor,
        )
        vertical_bar_charts, x_variables_selections = generate_vertical_bar_charts(
            base=base,
            vertical_bar_charts_config=vertical_bar_charts_config,
            predictor=predictor,
        )

        selections = dict(
            y_variables_selections,
            **x_variables_selections,
        )
        selection_charts = dict(
            horizontal_bar_charts,
            **vertical_bar_charts,
        )
        chart = add_interactive_selection(
            base=chart, selection_charts=selection_charts, selections=selections
        )

        chart = concatenate_charts(
            main_chart=chart,
            horizontal_bar_charts=horizontal_bar_charts,
            vertical_bar_charts=vertical_bar_charts,
            spacing=0,
        )
    elif "care_site_level" in estimates.columns:
        care_site_level_dropdwon = alt.binding_select(
            options=estimates["care_site_level"].unique(), name="Care site level : "
        )
        care_site_level_selection = alt.selection_point(
            fields=["care_site_level"],
            bind=care_site_level_dropdwon,
            value=estimates["care_site_level"].unique()[0],
        )
        chart = chart.add_params(care_site_level_selection).transform_filter(
            care_site_level_selection
        )
    if chart_style:
        chart = configure_style(chart=chart, chart_style=chart_style)
    if save_path:
        save_html(
            obj=chart,
            filename=save_path,
        )

    return chart