# Source: https://github.com/cwerner/covid19
# 1. Source data directly from GitHub (JHU COVID)
# 2. Configurable UI based on two variables - inhabitants and countries
# 3. Use Altair (https://altair-viz.github.io/) declarative statistical visualization charts
import datetime
from functools import reduce
import streamlit as st
from streamlit import caching
import pandas as pd
import altair as alt
import os
# numbers for 2019
inhabitants = {'India': 1352.6,
'US': 328.2,
'Brazil': 209.5,
'Russia': 144.5,
'United Kingdom': 67.1,
'China': 1392.7,
'Italy': 60.23}
@st.cache
def read_data():
BASEURL = "https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series"
url_confirmed = f"{BASEURL}/time_series_covid19_confirmed_global.csv"
url_deaths = f"{BASEURL}/time_series_covid19_deaths_global.csv"
url_recovered = f"{BASEURL}/time_series_covid19_recovered_global.csv"
confirmed = pd.read_csv(url_confirmed, index_col=0)
deaths = pd.read_csv(url_deaths, index_col=0)
recovered = pd.read_csv(url_recovered, index_col=0)
# sum over potentially duplicate rows (France and their territories)
confirmed = confirmed.groupby("Country/Region").sum().reset_index()
deaths = deaths.groupby("Country/Region").sum().reset_index()
recovered = recovered.groupby("Country/Region").sum().reset_index()
return (confirmed, deaths, recovered)
def transform(df, collabel='confirmed'):
dfm = pd.melt(df)
dfm["date"] = pd.to_datetime(dfm.variable, infer_datetime_format=True)
dfm = dfm.set_index("date")
dfm = dfm[["value"]]
dfm.columns = [collabel]
return dfm
def transform2(df, collabel='confirmed'):
dfm = pd.melt(df, id_vars=["Country/Region"])
dfm["date"] = pd.to_datetime(dfm.variable, infer_datetime_format=True)
dfm = dfm.set_index("date")
dfm = dfm[["Country/Region","value"]]
dfm.columns = ["country", collabel]
return dfm
def app():
st.title("🦠 Covid-19 Data Explorer")
st.markdown("""\
This app illustrates the spread of COVID-19 in select countries over time.
""")
#st.error("⚠️ There is currently an issue in the datasource of JHU. Data for 03/13 is invalid and thus removed!")
countries = ["India", "US", "Russia", "Brazil", "China", "Italy", "United Kingdom"]
analysis = st.sidebar.selectbox("Choose Analysis", ["Overview", "By Country"])
if analysis == "Overview":
st.header("COVID-19 cases and fatality rate")
st.markdown("""\
These are the reported case numbers for a selection of countries"""
f""" (currently only {', '.join(countries)}). """
"""The case fatality rate (CFR) is calculated as:
$$
CFR[\%] = \\frac{fatalities}{\\textit{all cases}}
$$
ℹ️ You can select/ deselect countries and switch between linear and log scales.
""")
confirmed, deaths, recovered = read_data()
multiselection = st.multiselect("Select countries:", countries, default=countries)
logscale = st.checkbox("Log scale", False)
confirmed = confirmed[confirmed["Country/Region"].isin(multiselection)]
confirmed = confirmed.drop(["Lat", "Long"],axis=1)
confirmed = transform2(confirmed, collabel="confirmed")
deaths = deaths[deaths["Country/Region"].isin(multiselection)]
deaths = deaths.drop(["Lat", "Long"],axis=1)
deaths = transform2(deaths, collabel="deaths")
frate = confirmed[["country"]]
frate["frate"] = (deaths.deaths / confirmed.confirmed)*100
# saveguard for empty selection
if len(multiselection) == 0:
return
SCALE = alt.Scale(type='linear')
if logscale:
confirmed["confirmed"] += 0.00001
confirmed = confirmed[confirmed.index > '2020-02-16']
frate = frate[frate.index > '2020-02-16']
SCALE = alt.Scale(type='log', domain=[10, int(max(confirmed.confirmed))], clamp=True)
c2 = alt.Chart(confirmed.reset_index()).properties(height=150).mark_line().encode(
x=alt.X("date:T", title="Date"),
y=alt.Y("confirmed:Q", title="Cases", scale=SCALE),
color=alt.Color('country:N', title="Country")
)
# case fatality rate...
c3 = alt.Chart(frate.reset_index()).properties(height=100).mark_line().encode(
x=alt.X("date:T", title="Date"),
y=alt.Y("frate:Q", title="Fatality rate [%]", scale=alt.Scale(type='linear')),
color=alt.Color('country:N', title="Country")
)
per100k = confirmed.loc[[confirmed.index.max()]].copy()
per100k.loc[:,'inhabitants'] = per100k.apply(lambda x: inhabitants[x['country']], axis=1)
per100k.loc[:,'per100k'] = per100k.confirmed / (per100k.inhabitants * 1_000_000) * 100_000
per100k = per100k.set_index("country")
per100k = per100k.sort_values(ascending=False, by='per100k')
per100k.loc[:,'per100k'] = per100k.per100k.round(2)
c4 = alt.Chart(per100k.reset_index()).properties(width=75).mark_bar().encode(
x=alt.X("per100k:Q", title="Cases per 100k inhabitants"),
y=alt.Y("country:N", title="Countries", sort=None),
color=alt.Color('country:N', title="Country"),
tooltip=[alt.Tooltip('country:N', title='Country'),
alt.Tooltip('per100k:Q', title='Cases per 100k'),
alt.Tooltip('inhabitants:Q', title='Inhabitants [mio]')]
)
st.altair_chart(alt.hconcat(c4, alt.vconcat(c2, c3)), use_container_width=True)
st.markdown(f"""\
⚠️ Please take the CFR with a grain of salt. The ratio is
highly dependend on the total number of tests conducted in a country. In the early stages
of the outbreak often mainly severe cases with clear symptoms are detected. Thus mild cases
are not recorded which skews the CFR.
""", unsafe_allow_html=True)
elif analysis == "By Country":
confirmed, deaths, recovered = read_data()
st.header("Country statistics")
st.markdown("""\
The reported number of active, recovered and deceased COVID-19 cases by country """
f""" (currently only {', '.join(countries)}).
"""
"""
ℹ️ You can select countries and plot data as cummulative counts or new active cases per day.
""")
# selections
selection = st.selectbox("Select country:", countries)
cummulative = st.radio("Display type:", ["total", "new cases"])
#scaletransform = st.radio("Plot y-axis", ["linear", "pow"])
confirmed = confirmed[confirmed["Country/Region"] == selection].iloc[:,3:]
confirmed = transform(confirmed, collabel="confirmed")
deaths = deaths[deaths["Country/Region"] == selection].iloc[:,3:]
deaths = transform(deaths, collabel="deaths")
recovered = recovered[recovered["Country/Region"] == selection].iloc[:,3:]
recovered = transform(recovered, collabel="recovered")
df = reduce(lambda a,b: pd.merge(a,b, on='date'), [confirmed, recovered, deaths])
df["active"] = df.confirmed - (df.deaths + df.recovered)
variables = ["recovered", "active", "deaths"]
colors = ["steelblue", "orange", "black"]
value_vars = variables
SCALE = alt.Scale(domain=variables, range=colors)
if cummulative == 'new cases':
value_vars = ["new"]
df["new"] = df.confirmed - df.shift(1).confirmed
df["new"].loc[df.new < 0] = 0
SCALE = alt.Scale(domain=["new"], range=["orange"])
dfm = pd.melt(df.reset_index(), id_vars=["date"], value_vars=value_vars)
# introduce order col as altair does auto-sort on stacked elements
dfm['order'] = dfm['variable'].replace(
{val: i for i, val in enumerate(variables[::-1])}
)
c = alt.Chart(dfm.reset_index()).mark_bar().properties(height=200).encode(
x=alt.X("date:T", title="Date"),
y=alt.Y("sum(value):Q", title="Cases", scale=alt.Scale(type='linear')),
color=alt.Color('variable:N', title="Category", scale=SCALE), #, sort=alt.EncodingSortField('value', order='ascending')),
order='order'
)
if cummulative != 'new cases':
st.altair_chart(c, use_container_width=True)
else:
# add smooth 7-day trend
rm_7day = df[['new']].rolling('7D').mean().rename(columns={'new': 'value'})
c_7day = alt.Chart(rm_7day.reset_index()).properties(height=200).mark_line(strokeDash=[1,1], color='red').encode(
x=alt.X("date:T", title="Date"),
y=alt.Y("value:Q", title="Cases", scale=alt.Scale(type='linear')),
)
st.altair_chart((c + c_7day), use_container_width=True)
st.markdown(f"""\
Daily reported new cases (incl. 7-day average).
""", unsafe_allow_html=True)
st.info("""\
by: [C. Werner](https://www.christianwerner.net) | source: [GitHub](https://www.github.com/cwerner/covid19)
| data source: [Johns Hopkins Univerity (GitHub)](https://github.com/CSSEGISandData/COVID-19).
""")
# ----------------------
app()