Plot covid incidence / death numbers for selected countries (austria+neighbors+uk)

at-covid19.py

#!/usr/bin/env python3
import pandas as pd
import matplotlib.pyplot as plt
import countryinfo  # for populations

FUTURE = 28
POPULATION = {
    "Czechia": countryinfo.CountryInfo("Czech Republic").population(),
}


NORMALIZE_TO = 100000
WINDOW_DAYS = 7

URL = "https://github.com/CSSEGISandData/COVID-19/raw/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv"
URL_DEATHS = "https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv"


if __name__ == "__main__":
    covid19 = pd.read_csv(URL)
    covid19_deaths = pd.read_csv(URL_DEATHS)
    # no provinces
    covid19 = covid19[covid19["Province/State"].isnull()]
    covid19_deaths = covid19_deaths[covid19_deaths["Province/State"].isnull()]

    def prepare_data(covid19, region):
        at_c19 = covid19[covid19["Country/Region"] == region]
        # the first four columns 'Province/State', 'Country/Region', 'Lat', 'Long' followed by actual data points, so we skip these.
        at_cases = at_c19[at_c19.columns[4:]]
        # in rows...
        at_cases_rows = at_cases.T
        new_cases = at_cases_rows.diff()

        # 7 day incidence
        if region not in POPULATION:
            POPULATION[region] = countryinfo.CountryInfo(region).population()
            print(f"population {region}: {POPULATION[region]}")
        seven_day_incidence = (new_cases / POPULATION[region] * NORMALIZE_TO).rolling(window=WINDOW_DAYS).sum()
        # we use the latest data point as label in the legend so we need the value
        last_value = seven_day_incidence.iloc[-1].tolist()[0]
        # and the label is generated from the column name so we change that:
        data_renamed = seven_day_incidence.rename(
            columns={
                seven_day_incidence.columns[0]: "%s %3.1f"
                % (
                    region,
                    last_value,
                )
            }
        )
        return data_renamed, new_cases, last_value

    def plot_for_region(region="Austria", axes=None):
        data_renamed, new_cases, last_value = prepare_data(covid19, region)
        death_renamed, new_death, last_death_value = prepare_data(covid19_deaths, region)

        axes = data_renamed.plot(
            logy=True,  # in the logy scale the growth is linear. use your ruler to predict the future :-/
            grid=True,
            xticks=range(0, len(new_cases) + FUTURE, WINDOW_DAYS),  # get a label for one day per week on the x-axis
            label=str(last_value),
            ax=axes,
        )
        line_color = axes.get_lines()[-1].get_color()

        axes = death_renamed.plot(
            logy=True,  # in the logy scale the growth is linear. use your ruler to predict the future :-/
            grid=True,
            xticks=range(0, len(new_death) + FUTURE, WINDOW_DAYS),  # get a label for one day per week on the x-axis
            label=str(last_value),
            ax=axes,
            linestyle="dashed",
            color=line_color,
        )
        line_color = axes.get_lines()[-1].get_color()

        return len(new_cases), line_color

    plt.close()
    axes = plt.subplot(111)
    plt.title("%s day incidence per %d" % (WINDOW_DAYS, NORMALIZE_TO))
    num_days = plot_for_region("Austria", axes=axes)
    for country in ["Germany", "Switzerland", "Liechtenstein", "Italy", "Slovenia", "Hungary", "Slovakia", "Czechia"]:
        _, lines = plot_for_region(country, axes=axes)
    plot_for_region("United Kingdom", axes=axes)
    axes.set_ylim(0.01, axes.get_ylim()[1])
    plt.xticks(rotation="vertical")
    plt.show()