Verwende die `full-data.csv` von pavelmayer.de/covid/risks/ um die Verteilung der Inzidenzen der Landkreise im zeitlichen Verlauf anhand ihrer Größe darzustellen. (Achtung, sehr schlechter Code...)

inzidenz_per_person.py

# -*- coding: utf-8 -*-

import pandas as pd
import numpy as np
import datetime



columns = [
    'IdBundesland',
    'Bundesland',
    'IdLandkreis',
    'Landkreis',
    'Meldedatum',
    'AnzahlFall',
    'Bevoelkerung',
    'Refdatum',
    'Datenstand',
]

utctimestamparser = lambda x: datetime.datetime.utcfromtimestamp(int(x)/1000)

df = pd.read_csv('full-data.csv', usecols = columns, parse_dates=['Meldedatum', 'Refdatum'], date_parser=utctimestamparser)

bevoelkerung = df[['IdLandkreis','Bevoelkerung']].groupby(['IdLandkreis']).first()['Bevoelkerung']


df["Datenstand"] = pd.to_datetime(df['Datenstand'],format="%d.%m.%Y, %H:%S Uhr")

dff = df[['IdBundesland','IdLandkreis','AnzahlFall','Meldedatum']]
dff = dff[dff['AnzahlFall'] > 0]

dff = dff.set_index(['IdLandkreis', 'Meldedatum'])['AnzahlFall']
dff = dff.groupby(level='IdLandkreis').resample('1d',level='Meldedatum').sum()

dff = dff.reindex(pd.MultiIndex.from_product([dff.index.levels[0],dff.index.levels[1]],names=['IdLandkreis','Meldedatum']), fill_value=0)

sum_7day = dff.groupby(level='IdLandkreis').apply(lambda x: x.rolling(7).sum())

inzidenz = (sum_7day / bevoelkerung * 100000).dropna()

inzidenz = inzidenz.swaplevel(0,1)

inzidenz = inzidenz.dropna().sort_index().rename('Inzidenz')
inzidenz = inzidenz.to_frame()

inzidenz = inzidenz.sort_values(['Meldedatum', 'Inzidenz'])

import matplotlib
import matplotlib.pyplot as plt
from matplotlib.collections import PatchCollection
from matplotlib.patches import Rectangle
rectangles = [];
colors = [];
breakpointx = []
breakpoints = [3, 7, 10, 20, 35, 50, 100, 200, 500]
breakpointpoints = [[] for i in range(len(breakpoints))]
for start in inzidenz.index.get_level_values('Meldedatum').unique():
    #width = matplotlib.dates.date2num(end) - matplotlib.dates.date2num(start)
    width = 1
    height = 0
    breakpointindex = 0
    breakpointheights = []
    breakpointx.append(matplotlib.dates.date2num(start))
    for i, (kreis, inz) in enumerate(inzidenz.loc[start, :].iterrows()):
        while breakpointindex < len(breakpoints) and inz[0] > breakpoints[breakpointindex]:
            breakpointheights.append(height)
            breakpointindex += 1
        rectangles.append(Rectangle((matplotlib.dates.date2num(start), height), width, bevoelkerung[kreis]))
        height += bevoelkerung[kreis]
        colors.append(inz[0])
    while breakpointindex < len(breakpoints):
        breakpointheights.append(height)
        breakpointindex += 1
    for i, height in enumerate(breakpointheights):
        breakpointpoints[i].append(height)

import copy
cmap = copy.copy(matplotlib.cm.turbo)
cmap.set_bad(cmap.colors[0])

fig = plt.figure(figsize=(8,4))
ax = fig.add_subplot(111)
p = PatchCollection(rectangles, cmap=cmap,
    norm=matplotlib.colors.LogNorm()
)
p.set_array(np.array(colors))
ax.add_collection(p)
cb =  plt.colorbar(p)
cb.set_label('7-Tage Inzidenz anhand Meldedatum', rotation=270)


import matplotlib.units as munits
import matplotlib.dates as mdates
converter = mdates.ConciseDateConverter()
munits.registry[np.datetime64] = converter
munits.registry[datetime.date] = converter
munits.registry[datetime.datetime] = converter

# assign date locator / formatter to the x-axis to get proper labels
locator = mdates.MonthLocator()
formatter = mdates.ConciseDateFormatter(locator)
ax.xaxis.set_major_locator(locator)
ax.xaxis.set_major_formatter(formatter)
ax.xaxis.set_minor_locator(mdates.WeekdayLocator(matplotlib.dates.MO))
for i, breakpoint in enumerate(breakpoints):
    plt.plot(breakpointx, breakpointpoints[i], label=f"{breakpoint}")

plt.xlim([
    matplotlib.dates.date2num(inzidenz.index.levels[0].min()),
    matplotlib.dates.date2num(inzidenz.index.levels[0].max())
    ])
plt.ylim([0, bevoelkerung.sum()])
plt.title("Inzidenz pro Person in Landkreisen")
plt.ylabel("Personen in Landkreisen")
plt.xlabel("Meldedatum")
plt.legend()
#plt.show()
plt.savefig('./output.png', dpi=400)

Ausgabe: