A simple Python script to generate an Edward-Tufte-style, small multiples graph of the case rate (7-day average) by State for the U.S. See the sample output in this blog post: https://sprestridge.net/datavis/python/covid-19/2020/06/05/COVID-Small-Multiples.html

covid19-small-multiples.py

# covid19-small-multiples

import math
import pandas as pd
import numpy as np
import seaborn as sns

# setup for plotting
import matplotlib as mpl
from matplotlib import pyplot as plt
from matplotlib.colors import ListedColormap
from matplotlib.patches import Patch

from IPython.display import HTML
from IPython.display import clear_output

from datetime import datetime

plt.style.use('fivethirtyeight')

#%matplotlib inline
#%config InlineBackend.figure_format = 'retina'

# Increase default figure and font sizes for easier viewing.
plt.rcParams['figure.figsize'] = (8, 6)
plt.rcParams['font.size'] = 14

# define a custom function for rounding up
def round_up(n, decimals=0):
    multiplier = 10 ** decimals
    return math.ceil(n * multiplier) / multiplier


# import some data - this is from the NY Times
url = 'https://raw.githubusercontent.com/nytimes/covid-19-data/master/us-states.csv'
us_states = pd.read_csv(url, parse_dates=['date'])

# Replaces spaces with "-" and make column names lowercase
# NOTE: combine into lowercase and underscore w/ one line!!
us_states.columns = us_states.columns.str.lower().str.replace(" ", "-")

# put the states in a unique list
list_states = us_states['state'].sort_values().unique()

# most recent date in the dataframe
mydate = us_states.date.max()

# convert to a string for use in filename later
str_mydate = mydate.strftime("%Y-%m-%d")


# SMALL MULTIPLES - All States

# surpress SettingWithCopyWarning
pd.set_option('mode.chained_assignment',None)

# delete the states that won't have enough data to plot
list_states = list_states[(list_states != 'Alaska') & (list_states != 'Hawaii') & (list_states != 'Montana') 
                         & (list_states != 'Northern Mariana Islands') & (list_states != 'Virgin Islands')
                         & (list_states != 'Wyoming')]

num = 0
ncols = 5
nrows = int(np.ceil(len(list_states) / ncols))

fig, axes = plt.subplots(nrows = nrows, ncols = ncols, figsize = (20, nrows*3))

# loop through states and plot
for state in list_states:
    df = us_states[(us_states.state == state)]

    # add calculated cases per day column (NOTE: SettingWithCopyWarning)
    df['cases_per_day'] = df.cases.diff()

    # add calculated case-rate column (NOTE: SettingWithCopyWarning)
    df['case_rate'] = np.round(df.cases_per_day.rolling(window=7).mean(),0)

    # mask for case_rate >= 30 days
    df = df[(df.case_rate >= 30)]

    # maximum y-axis, calculated
    max_y = round_up(df.case_rate.max(), -2)

    # interval for y-axis
    interval_y = max_y / 4

    # find the right spot on the plot
    num+=1
    plt.subplot(nrows, ncols, num)
    
    ax = df.case_rate.plot(kind="line", xticks = (),
                  title = state,
                  ylim = (0, max_y), yticks=(0, interval_y, 2*interval_y, 3*interval_y, 4*interval_y));

    ax.yaxis.set_major_formatter(mpl.ticker.StrMethodFormatter('{x:,.0f}'))

fig.tight_layout()
fig.set_facecolor('w')
plt.suptitle("Case-rate by state (rolling 7-day ave.) - " + str_mydate + " \n", fontsize=32, 
             fontweight=0, color='black', style='oblique')
plt.xlabel('Source: NY Times COVID-19 Data')
plt.subplots_adjust(top=0.95)

str_filename = 'case-rate-small-multiples-' + str_mydate + '.png'
fig.savefig(str_filename);