Our World in Data - Compare Covid Data across Years

owid-covid.py

#!/usr/bin/env python3
import sys
import matplotlib.pyplot as plt
import requests as rq
import pandas as pd
import numpy as np
from typing import cast
from io import StringIO
from datetime import date

def usage():
    print(f"usage: covid.py [--help|--list-columns] <country> [column='{DEFAULT_COLUMN}']")
    sys.exit(1)

DEFAULT_COLUMN = "new_cases_smoothed"

args = sys.argv[1:]
action = "country"
while len(args) > 0 and args[0].startswith("-"):
    if args[0] == "--help":
        usage()
    elif args[0] == "--list-columns":
        action = "list-columns"
    else:
        print(f"error: invalid option '{args[0]}'")
        sys.exit(1)
    args = args[1:]

if action == "country" and len(args) != 1 and len(args) != 2:
    usage()
elif action == "list-columns" and len(args) != 0:
    usage()

print(">> Downloading Covid Data")
r = rq.get("https://covid.ourworldindata.org/data/owid-covid-data.csv")
if r.status_code != 200:
    raise ValueError(f"HTTP error {r.status_code}")

print(">> Parsing CSV")
df = cast(pd.DataFrame, pd.read_csv(StringIO(r.text)))

if action == "country":
    country = args[0]
    if len(args) == 2:
        column = args[1]
    else:
        column = DEFAULT_COLUMN

    if country not in df.location.iloc:
        print(f"error: invalid country '{country}'")
        sys.exit(1)

    df = df[df.location == country]

    print(">> Collecting years with data")
    dates = [date.fromisoformat(datestr) for datestr in df.date.iloc]
    years = list(set(date.year for date in dates))
    print(years)

    print(">> Plotting year data")
    for year in years:
        year_start = date(year, 1, 1)
        year_end = date(year, 12, 31)
        year_dates = [(dates[i] - year_start).days for i in range(len(dates)) if year_start <= dates[i] <= year_end]
        year_values = [df[column].iloc[i] for i in range(len(dates)) if year_start <= dates[i] <= year_end]

        plt.plot(year_dates, year_values, label = year)

    plt.title(f"{country} - {column}")
    plt.xticks(
        [(date(2000, i, 1) - date(2000, 1, 1)).days for i in range(1, 13)],
        [i for i in range(1, 13)],
    )
    plt.legend()
    plt.show()
elif action == "list-columns":
    print(">> List of Columns")
    for column in sorted(df.columns):
        if df.dtypes[column].type not in [np.float64, np.uint64]:
            continue
        print(column)

Example usage

$ python owid-covid.py Austria

$ python owid-covid.py Austria icu_patients