How to adjust tooltip to display only column selected by slider in Bokeh-CodePudding

I am trying to crate visualization using Geopandas and Bokeh. Data has column region, geometry and >100 columns (week1, week2, week3,..) that have numeric value of cases.

The task is for the tooltip when hovering over a region to show only the value from column currently selected by the slider.

Things I tried:

This option displays all columns in the tooltip. Too much information

data.plot_bokeh(
slider=[week1,week2,week3],
slider_name="Week:",
figsize=(1600,600))

Not sure what to put as variable next to "Cases", or how to make it dynamic based on slider.

data.plot_bokeh(
slider=[week1,week2,week3],
slider_name="Week:",
hovertool_string="""<h1>@region_name</h1>
                    <h3>Cases: ?? </h3>""",
figsize=(1600,600))

Bokeh has to have the current value according to slider stored somewhere, as it's correctly coloring the region based on value

CodePudding user response：

if you inspect the code of plotly_bokeh in geoplot.py/geoplot() you will find in slider callback JavaScript an internal column Colormap is set to column corresponding to value in slider (code below)
you did not provide a MWE, so have used OWID COVID data and natural earth lowres polygons. (see data preparation)
with this it is simple to define hovertool_string argument to work the way you want

solution


import warnings

# bokeh generates a lot of shapely deprecation warnings
with warnings.catch_warnings():
    warnings.simplefilter("ignore")

    gdf.plot_bokeh(
        slider=[c for c in gdf.columns if c[0:4] == "week"],
        slider_name="week ",
        colorbar_tick_format="0.0a",
        colormap="Inferno",
        colormap_range=[
            gdf.loc[:, [c for c in gdf.columns if c[0:4] == "week"]].quantile(p).quantile(p)
            for p in [0.10, 0.90]
        ],
        hovertool_string="<h3>@name</h3><pre>@Colormap{0.0a}</pre>"
    )

output

plotly_bokeh callback

        # Define Callback for Slider widget:
        callback = CustomJS(
            args=dict(
                slider_widget=slider_widget,
                geo_source=geo_source,
                value2name=value2name,
            ),
            code="""

                //Change selection of field for Colormapper for choropleth plot:
                var slider_value = slider_widget.value;
                var i;
                for(i=0; i<value2name.data["Names"].length; i  )
                    {
                    if (value2name.data["Values"][i] == slider_value)
                        {
                         var name = value2name.data["Names"][i];
                         }

                    }
                geo_source.data["Colormap"] = geo_source.data[name];
                geo_source.change.emit();

                            """,
        )
        slider_widget.js_on_change("value", callback)

data preparation

import pandas as pd
import geopandas as gpd
import pandas_bokeh
pandas_bokeh.output_notebook()

df = pd.read_csv(
    "https://raw.githubusercontent.com/owid/covid-19-data/master/public/data/owid-covid-data.csv"
).loc[
    lambda df: ~df["iso_code"].str.contains("_"),
    ["iso_code", "date", "total_cases_per_million"],
]
df["date"] = pd.to_datetime(df["date"])

# cases for each week as columns
df2 = (
    df.merge(
        pd.DataFrame(
            {"date": pd.date_range(df["date"].quantile(.1), df["date"].max(), freq="W-MON")}
        ).assign(week=lambda d: "week"   d.index.astype(str).str.zfill(3)),
        on="date",
        how="inner",
    )
    .drop(columns=["date"])
    .set_index(["iso_code", "week"])
    .unstack("week")
    .droplevel(0, 1)
    .reset_index()
)

# geometry and cases as columns
gdf = (
    gpd.read_file(gpd.datasets.get_path("naturalearth_lowres"))
    .loc[:, ["iso_a3", "name", "geometry"]]
    .merge(df2, left_on="iso_a3", right_on="iso_code")
)