edsteva.viz.dashboards

probe_dashboard

probe_dashboard(
    probe: BaseProbe,
    fitted_model: BaseModel = None,
    care_site_level: List[str] = None,
    save_path: str = None,
    legend_predictor: str = "Predictor c(t)",
    legend_model: str = "Model f(t)",
    x_axis_title: str = None,
    y_axis_title: str = None,
    main_chart_config: Dict[str, float] = None,
    model_line_config: Dict[str, str] = None,
    probe_line_config: Dict[str, str] = None,
    vertical_bar_charts_config: Dict[str, str] = None,
    horizontal_bar_charts_config: Dict[str, str] = None,
    time_line_config: Dict[str, str] = None,
    chart_style: Dict[str, float] = None,
    **kwargs
)

Displays an interactive chart with:

• On the top, the aggregated average completeness predictor \(c(t)\) over time \(t\) with the fitted model \(\hat{c}(t)\) if specified.
• On the bottom, interactive filters including all the concepts in the Probe (such as time, care site, number of visits...etc.)

Is is possible to save the chart in HTML with the "save_path" optional input.

PARAMETER	DESCRIPTION
probe	Class describing the completeness predictor \(c(t)\). TYPE: BaseProbe
fitted_model	Model fitted to the probe. TYPE: BaseModel DEFAULT: None
care_site_level	EXAMPLE: ["Hospital"], ["Hôpital", "UF"] or ["UF", "UH"] TYPE: List[str] DEFAULT: None
save_path	Folder path where to save the chart in HTML format. EXAMPLE: "my_folder/my_file.html" TYPE: str DEFAULT: None
legend_predictor	Label name for the predictor legend. TYPE: str DEFAULT: 'Predictor c(t)'
legend_model	Label name for the model legend. TYPE: str DEFAULT: 'Model f(t)'
x_axis_title	Label name for the x axis. TYPE: str DEFAULT: None
y_axis_title	Label name for the y axis. TYPE: str DEFAULT: None
main_chart_config	If not None, configuration used to construct the top main chart. TYPE: Dict[str, float] DEFAULT: None
model_line_config	If not None, configuration used to construct the model line. TYPE: Dict[str, str] DEFAULT: None
probe_line_config	If not None, configuration used to construct the probe line. TYPE: Dict[str, str] DEFAULT: None
vertical_bar_charts_config	If not None, configuration used to construct the vertical bar charts. TYPE: Dict[str, str] DEFAULT: None
horizontal_bar_charts_config	If not None, configuration used to construct the horizontal bar charts. TYPE: Dict[str, str] DEFAULT: None
time_line_config	If not None, configuration used to construct the time line. TYPE: Dict[str, str] DEFAULT: None
chart_style	If not None, configuration used to configure the chart style. EXAMPLE: {"labelFontSize": 13, "titleFontSize": 14} TYPE: Dict[str, float] DEFAULT: None

Source code in edsteva/viz/dashboards/probe/wrapper.py

def probe_dashboard(
    probe: BaseProbe,
    fitted_model: BaseModel = None,
    care_site_level: List[str] = None,
    save_path: str = None,
    legend_predictor: str = "Predictor c(t)",
    legend_model: str = "Model f(t)",
    x_axis_title: str = None,
    y_axis_title: str = None,
    main_chart_config: Dict[str, float] = None,
    model_line_config: Dict[str, str] = None,
    probe_line_config: Dict[str, str] = None,
    vertical_bar_charts_config: Dict[str, str] = None,
    horizontal_bar_charts_config: Dict[str, str] = None,
    time_line_config: Dict[str, str] = None,
    chart_style: Dict[str, float] = None,
    **kwargs,
):
    r"""Displays an interactive chart with:

    - On the top, the aggregated average completeness predictor $c(t)$ over time $t$ with the fitted model $\hat{c}(t)$ if specified.
    - On the bottom, interactive filters including all the concepts in the [Probe][probe] (such as time, care site, number of visits...etc.)

    Is is possible to save the chart in HTML with the "save_path" optional input.

    Parameters
    ----------
    probe : BaseProbe
        Class describing the completeness predictor $c(t)$.
    fitted_model : BaseModel, optional
        Model fitted to the probe.
    care_site_level : List[str], optional
        **EXAMPLE**: `["Hospital"]`, `["Hôpital", "UF"]` or `["UF", "UH"]`
    save_path : str, optional
        Folder path where to save the chart in HTML format.

        **EXAMPLE**: `"my_folder/my_file.html"`
    legend_predictor: str, optional,
        Label name for the predictor legend.
    legend_model: str, optional,
        Label name for the model legend.
    x_axis_title: str, optional,
        Label name for the x axis.
    y_axis_title: str, optional,
        Label name for the y axis.
    main_chart_config: Dict[str, str], optional
        If not None, configuration used to construct the top main chart.
    model_line_config: Dict[str, str], optional
        If not None, configuration used to construct the model line.
    probe_line_config: Dict[str, str], optional
        If not None, configuration used to construct the probe line.
    vertical_bar_charts_config: Dict[str, str], optional
        If not None, configuration used to construct the vertical bar charts.
    horizontal_bar_charts_config: Dict[str, str], optional
        If not None, configuration used to construct the horizontal bar charts.
    time_line_config: Dict[str, str], optional
        If not None, configuration used to construct the time line.
    chart_style: Dict[str, float], optional
        If not None, configuration used to configure the chart style.

        **EXAMPLE**: `{"labelFontSize": 13, "titleFontSize": 14}`
    """

    alt.data_transformers.enable("default")
    alt.data_transformers.disable_max_rows()

    probe_config = deepcopy(probe.get_viz_config("probe_dashboard"))
    if fitted_model:
        model_config = deepcopy(fitted_model.get_viz_config("probe_dashboard"))
        if model_line_config is None:
            model_line_config = model_config["model_line"]
        if probe_line_config is None:
            probe_line_config = model_config["probe_line"]
    if main_chart_config is None:
        main_chart_config = probe_config["main_chart"]
    if time_line_config is None:
        time_line_config = probe_config["time_line"]
    if vertical_bar_charts_config is None:
        vertical_bar_charts_config = probe_config["vertical_bar_charts"]
        if fitted_model:
            vertical_bar_charts_config["y"] = (
                vertical_bar_charts_config["y"]
                + model_config["extra_vertical_bar_charts"]
            )
    if horizontal_bar_charts_config is None:
        horizontal_bar_charts_config = probe_config["horizontal_bar_charts"]
        if fitted_model:
            horizontal_bar_charts_config["x"] = (
                horizontal_bar_charts_config["x"]
                + model_config["extra_horizontal_bar_charts"]
            )
    if chart_style is None:
        chart_style = probe_config["chart_style"]

    predictor = fitted_model.predict(probe) if fitted_model else probe.predictor.copy()
    predictor = filter_data(
        data=predictor,
        care_site_level=care_site_level,
        **kwargs,
    )

    if fitted_model:
        chart = fitted_probe_dashboard(
            predictor=predictor,
            legend_predictor=legend_predictor,
            legend_model=legend_model,
            x_axis_title=x_axis_title,
            y_axis_title=y_axis_title,
            main_chart_config=main_chart_config,
            model_line_config=model_line_config,
            probe_line_config=probe_line_config,
            vertical_bar_charts_config=vertical_bar_charts_config,
            horizontal_bar_charts_config=horizontal_bar_charts_config,
            time_line_config=time_line_config,
            chart_style=chart_style,
        )
    else:
        chart = probe_only_dashboard(
            predictor=predictor,
            x_axis_title=x_axis_title,
            y_axis_title=y_axis_title,
            main_chart_config=main_chart_config,
            vertical_bar_charts_config=vertical_bar_charts_config,
            horizontal_bar_charts_config=horizontal_bar_charts_config,
            time_line_config=time_line_config,
            chart_style=chart_style,
        )

    vis_probe = "id" + uuid.uuid4().hex
    new_index_probe_id = "id" + uuid.uuid4().hex
    old_index_probe_id = "id" + uuid.uuid4().hex
    left_shift = "145px" if fitted_model else "45px"
    html_chart = f"""
        <!DOCTYPE html>
        <html>
        <head>
          <script src="https://cdn.jsdelivr.net/npm/vega@{alt.VEGA_VERSION}"></script>
          <script src="https://cdn.jsdelivr.net/npm/vega-lite@{alt.VEGALITE_VERSION}"></script>
          <script src="https://cdn.jsdelivr.net/npm/vega-embed@{alt.VEGAEMBED_VERSION}"></script>
        </head>
        <body>

        <div class="container">
          <div class="row">
            <div>
            <div id={vis_probe}></div>
            </div>
            <div style="position:absolute;left:{left_shift};top:380px;width: -webkit-fill-available;">
            <div id={new_index_probe_id}>
              <div id={old_index_probe_id}></div>
            </div>
            <hr/>
            <h1 style="text-align:center"> Interactive filters </h1>
            </div>
          </div>
        </div>

        <script type="text/javascript">
        vegaEmbed('#{vis_probe}', {chart.to_json(indent=None)}).then(function(result) {{
            const sliders = document.getElementsByClassName('vega-bindings');
            const newparent = document.getElementById('{new_index_probe_id}');
            const oldchild = document.getElementById('{old_index_probe_id}');
            for (var i = 0; i < sliders.length; i++) {{
                if (sliders[i].parentElement.parentElement.id == '{vis_probe}') {{
                    var index_slider = sliders[i]
                    }}
                }}
            newparent.replaceChild(index_slider, oldchild);
            }}).catch(console.error);
        </script>
        </body>
        </html>
        """
    if save_path:
        save_html(
            obj=html_chart,
            filename=save_path,
        )
    else:
        display(HTML(html_chart))

normalized_probe_dashboard

normalized_probe_dashboard(
    probe: BaseProbe,
    fitted_model: BaseModel,
    care_site_level: List[str] = None,
    save_path: str = None,
    x_axis_title: str = None,
    y_axis_title: str = None,
    main_chart_config: Dict[str, str] = None,
    model_line_config: Dict[str, str] = None,
    error_line_config: Dict[str, str] = None,
    probe_line_config: Dict[str, str] = None,
    estimates_selections: Dict[str, str] = None,
    estimates_filters: Dict[str, str] = None,
    vertical_bar_charts_config: Dict[str, str] = None,
    horizontal_bar_charts_config: Dict[str, str] = None,
    time_line_config: Dict[str, str] = None,
    chart_style: Dict[str, float] = None,
    indexes_to_remove: List[str] = ["care_site_level"],
    **kwargs
)

Displays an interactive chart with:

• On the top, the aggregated normalized completeness predictor \(\frac{c(\Delta t)}{c_0}\) over normalized time \(\Delta t = t - t_0\). It represents the overall deviation from the Model.
• On the bottom, interactive filters including all the columns in the Probe (such as time, care site, number of visits...etc.) and all the estimates (coefficients and metrics) in the Model.

Is is possible to save the chart in HTML with the "save_path" optional input.

PARAMETER	DESCRIPTION
probe	Class describing the completeness predictor \(c(t)\) TYPE: BaseProbe
fitted_model	Model fitted to the probe TYPE: BaseModel
care_site_level	EXAMPLE: ["Hospital"], ["Hôpital", "UF"] or ["UF", "UH"] TYPE: List[str] DEFAULT: None
save_path	Folder path where to save the chart in HTML format. TYPE: str DEFAULT: None
x_axis_title	Label name for the x axis. TYPE: str DEFAULT: None
y_axis_title	Label name for the y axis. TYPE: str DEFAULT: None
main_chart_config	If not None, configuration used to construct the top main chart. TYPE: Dict[str, str] DEFAULT: None
model_line_config	If not None, configuration used to construct the model line. TYPE: Dict[str, str] DEFAULT: None
error_line_config	If not None, configuration used to construct the error line. TYPE: Dict[str, str] DEFAULT: None
probe_line_config	If not None, configuration used to construct the probe line. TYPE: Dict[str, str] DEFAULT: None
estimates_selections	If not None, configuration used to construct the estimates selections. TYPE: Dict[str, str] DEFAULT: None
estimates_filters	If not None, configuration used to construct the estimates filters. TYPE: Dict[str, str] DEFAULT: None
vertical_bar_charts_config	If not None, configuration used to construct the vertical bar charts. TYPE: Dict[str, str] DEFAULT: None
horizontal_bar_charts_config	If not None, configuration used to construct the horizontal bar charts. TYPE: Dict[str, str] DEFAULT: None
time_line_config	If not None, configuration used to construct the time line. TYPE: Dict[str, str] DEFAULT: None
chart_style	If not None, configuration used to configure the chart style. EXAMPLE: {"labelFontSize": 13, "titleFontSize": 14} TYPE: Dict[str, float] DEFAULT: None
indexes_to_remove	indexes to remove from the groupby selection. TYPE: List[str] DEFAULT: ['care_site_level']

Source code in edsteva/viz/dashboards/normalized_probe/normalized_probe.py

def normalized_probe_dashboard(
    probe: BaseProbe,
    fitted_model: BaseModel,
    care_site_level: List[str] = None,
    save_path: str = None,
    x_axis_title: str = None,
    y_axis_title: str = None,
    main_chart_config: Dict[str, str] = None,
    model_line_config: Dict[str, str] = None,
    error_line_config: Dict[str, str] = None,
    probe_line_config: Dict[str, str] = None,
    estimates_selections: Dict[str, str] = None,
    estimates_filters: Dict[str, str] = None,
    vertical_bar_charts_config: Dict[str, str] = None,
    horizontal_bar_charts_config: Dict[str, str] = None,
    time_line_config: Dict[str, str] = None,
    chart_style: Dict[str, float] = None,
    indexes_to_remove: List[str] = ["care_site_level"],
    **kwargs,
):
    r"""Displays an interactive chart with:

    - On the top, the aggregated normalized completeness predictor $\frac{c(\Delta t)}{c_0}$ over normalized time $\Delta t = t - t_0$. It represents the overall deviation from the Model.
    - On the bottom, interactive filters including all the columns in the [Probe][probe] (such as time, care site, number of visits...etc.) and all the estimates (coefficients and metrics) in the [Model][model].

    Is is possible to save the chart in HTML with the "save_path" optional input.

    Parameters
    ----------
    probe : BaseProbe
        Class describing the completeness predictor $c(t)$
    fitted_model : BaseModel
        Model fitted to the probe
    care_site_level : List[str], optional
        **EXAMPLE**: `["Hospital"]`, `["Hôpital", "UF"]` or `["UF", "UH"]`
    save_path : str, optional
        Folder path where to save the chart in HTML format.
    x_axis_title: str, optional,
        Label name for the x axis.
    y_axis_title: str, optional,
        Label name for the y axis.
    main_chart_config: Dict[str, str], optional
        If not None, configuration used to construct the top main chart.
    model_line_config: Dict[str, str], optional
        If not None, configuration used to construct the model line.
    error_line_config: Dict[str, str], optional
        If not None, configuration used to construct the error line.
    probe_line_config: Dict[str, str], optional
        If not None, configuration used to construct the probe line.
    estimates_selections: Dict[str, str], optional
        If not None, configuration used to construct the estimates selections.
    estimates_filters: Dict[str, str], optional
        If not None, configuration used to construct the estimates filters.
    vertical_bar_charts_config: Dict[str, str], optional
        If not None, configuration used to construct the vertical bar charts.
    horizontal_bar_charts_config: Dict[str, str], optional
        If not None, configuration used to construct the horizontal bar charts.
    time_line_config: Dict[str, str], optional
        If not None, configuration used to construct the time line.
    chart_style: Dict[str, float], optional
        If not None, configuration used to configure the chart style.
        **EXAMPLE**: `{"labelFontSize": 13, "titleFontSize": 14}`
    indexes_to_remove: List[str], optional
        indexes to remove from the groupby selection.
    """
    alt.data_transformers.disable_max_rows()

    # Pre-processing
    predictor = probe.predictor.copy()
    estimates = fitted_model.estimates.copy()
    predictor_metrics = probe._metrics.copy()
    indexes = get_indexes_to_groupby(
        predictor_columns=predictor.columns,
        predictor_metrics=predictor_metrics,
        indexes_to_remove=indexes_to_remove,
    )
    predictor = predictor.merge(estimates, on=probe._index)
    predictor["normalized_date"] = month_diff(
        predictor["date"], predictor["t_0"]
    ).astype(int)
    for estimate in fitted_model._coefs + fitted_model._metrics:
        if pd.api.types.is_datetime64_any_dtype(predictor[estimate]):
            predictor[estimate] = predictor[estimate].dt.strftime("%Y-%m")
    predictor["normalized_c"] = predictor["c"].where(
        (predictor["normalized_date"] < 0) | (predictor["c_0"] == 0),
        predictor["c"] / predictor["c_0"],
    )
    predictor["model"] = 1
    predictor["model"] = predictor["model"].where(predictor["normalized_date"] >= 0, 0)
    predictor = filter_data(data=predictor, care_site_level=care_site_level, **kwargs)

    # Get viz config
    probe_config = deepcopy(probe.get_viz_config("normalized_probe_dashboard"))
    model_config = deepcopy(
        fitted_model.get_viz_config("normalized_probe_dashboard", predictor=predictor)
    )
    if main_chart_config is None:
        main_chart_config = probe_config["main_chart"]
    if time_line_config is None:
        time_line_config = probe_config["time_line"]
    if vertical_bar_charts_config is None:
        vertical_bar_charts_config = probe_config["vertical_bar_charts"]
        vertical_bar_charts_config["y"] = (
            vertical_bar_charts_config["y"] + model_config["extra_vertical_bar_charts"]
        )
    if horizontal_bar_charts_config is None:
        horizontal_bar_charts_config = probe_config["horizontal_bar_charts"]
        horizontal_bar_charts_config["x"] = (
            horizontal_bar_charts_config["x"]
            + model_config["extra_horizontal_bar_charts"]
        )
    if chart_style is None:
        chart_style = probe_config["chart_style"]
    if probe_line_config is None:
        probe_line_config = model_config["probe_line"]
    if model_line_config is None:
        model_line_config = model_config["model_line"]
    if error_line_config is None:
        error_line_config = model_config["error_line"]
    if estimates_selections is None:
        estimates_selections = model_config["estimates_selections"]
    if estimates_filters is None:
        estimates_filters = model_config["estimates_filters"]

    # Viz
    predictor["legend_model"] = (
        model_line_config.get("legend_title")
        if model_line_config.get("legend_title")
        else type(fitted_model).__name__
    )
    predictor["legend_predictor"] = probe_line_config["legend_title"]
    predictor["legend_error_band"] = error_line_config["legend_title"]
    base = alt.Chart(predictor)
    time_line, time_selection = generate_time_line(
        base=base,
        time_line_config=time_line_config,
    )

    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(
        date=time_selection,
        **y_variables_selections,
        **x_variables_selections,
    )
    selection_charts = dict(
        horizontal_bar_charts,
        **vertical_bar_charts,
    )
    base = add_interactive_selection(
        base=base,
        selection_charts=selection_charts,
        selections=selections,
    )
    base = add_estimates_filters(
        base=base,
        selection_charts=selection_charts,
        estimates_filters=estimates_filters,
    )
    index_selection, index_fields = create_groupby_selection(
        indexes=indexes,
        predictor=predictor,
    )
    main_chart = generate_main_chart(
        base=base,
        main_chart_config=main_chart_config,
        index_selection=index_selection,
        index_fields=index_fields,
        x_axis_title=x_axis_title,
        y_axis_title=y_axis_title,
    )
    probe_line = generate_probe_line(
        main_chart=main_chart, probe_line_config=probe_line_config
    )
    error_line = generate_error_line(
        main_chart=main_chart, error_line_config=error_line_config
    )
    model_line = generate_model_line(
        main_chart=main_chart, model_line_config=model_line_config
    )

    main_chart = probe_line + error_line + model_line
    if index_selection:
        main_chart = main_chart.add_params(index_selection)
    chart = concatenate_charts(
        main_chart=main_chart,
        time_line=time_line,
        horizontal_bar_charts=horizontal_bar_charts,
        vertical_bar_charts=vertical_bar_charts,
        spacing=0,
    )
    chart = configure_style(chart=chart, chart_style=chart_style)
    for estimate_selection in estimates_selections:
        chart = chart.add_params(estimate_selection)

    vis_threshold = "id" + uuid.uuid4().hex
    new_sliders_threshold_id = "id" + uuid.uuid4().hex
    old_sliders_threshold_id = "id" + uuid.uuid4().hex
    new_index_threshold_id = "id" + uuid.uuid4().hex
    old_index_threshold_id = "id" + uuid.uuid4().hex
    html_chart = f"""
        <!DOCTYPE html>
        <html>
        <head>
          <script src="https://cdn.jsdelivr.net/npm/vega@{alt.VEGA_VERSION}"></script>
          <script src="https://cdn.jsdelivr.net/npm/vega-lite@{alt.VEGALITE_VERSION}"></script>
          <script src="https://cdn.jsdelivr.net/npm/vega-embed@{alt.VEGAEMBED_VERSION}"></script>
        </head>
        <body>

        <div class="container">
          <div class="row">
            <div>
            <div id={vis_threshold}></div>
            </div>
            <div style="position:absolute;left:1000px;top:540px;width: -webkit-fill-available;">
            <div id={new_sliders_threshold_id}>
              <div id={old_sliders_threshold_id}></div>
            </div>
            </div>
            <div style="position:absolute;left:45px;top:410px;width: -webkit-fill-available;">
            <div id={new_index_threshold_id}>
              <div id={old_index_threshold_id}></div>
            </div>
            <hr/>
            <h1 style="text-align:center"> Interactive filters </h1>
            </div>
          </div>
        </div>

        <script type="text/javascript">
        vegaEmbed('#{vis_threshold}', {chart.to_json(indent=None)}).then(function(result) {{
            const sliders = document.getElementsByClassName('vega-bindings');
            const newestimate = document.getElementById('{new_sliders_threshold_id}');
            const oldestimate = document.getElementById('{old_sliders_threshold_id}');
            const newparent = document.getElementById('{new_index_threshold_id}');
            const oldchild = document.getElementById('{old_index_threshold_id}');
            for (var i = 0; i < sliders.length; i++) {{
                if (sliders[i].parentElement.parentElement.id == '{vis_threshold}') {{
                    var estimate_slider = sliders[i]
                    var index_slider = estimate_slider.querySelectorAll(".vega-bind")
                    }}
                }}
            newestimate.replaceChild(estimate_slider, oldestimate);
            for (var i = 0; i < index_slider.length; i++) {{
                if (index_slider[i].firstChild.innerHTML == "Group by: ") {{
                    var index_color = index_slider[i]}}
                }}
            newparent.replaceChild(index_color, oldchild);
            }}).catch(console.error);
        </script>
        </body>
        </html>
        """
    display(HTML(html_chart))
    if save_path:
        save_html(
            obj=html_chart,
            filename=save_path,
        )