notebook.community

Edit and run

Compare Country Trajectories - Death Rate

Comparing how countries death rate trajectories are similar with Italy, South Korea and Japan

  • comments: true
  • author: Pratap Vardhan
  • categories: [growth, compare, death, interactive]
  • image: images/covid-compare-country-death-trajectories.png
  • permalink: /compare-country-death-trajectories/




In [1]:



    


#hide
import pandas as pd
import altair as alt
from IPython.display import HTML



    











In [2]:



    


#hide
url = ('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/'
       'csse_covid_19_time_series/time_series_covid19_deaths_global.csv')
df = pd.read_csv(url)
# rename countries
df['Country/Region'] = df['Country/Region'].replace({'Korea, South': 'South Korea'})
df = df[~df['Country/Region'].isin(['Cruise Ship'])]   # Remove Ships
dt_cols = df.columns[~df.columns.isin(['Province/State', 'Country/Region', 'Lat', 'Long'])]



    











In [3]:



    


#hide
dff = (df.groupby('Country/Region')[dt_cols].sum()
       .stack().reset_index(name='Confirmed Cases')
       .rename(columns={'level_1': 'Date', 'Country/Region': 'Country'}))
dff['Date'] = pd.to_datetime(dff['Date'], format='%m/%d/%y')



    











In [4]:



    


#hide
MIN_CASES = 10
LAST_DATE = dt_cols[-1]
# sometimes last column may be empty, then go backwards
for c in dt_cols[::-1]:
    if not df[c].fillna(0).eq(0).all():
        LAST_DATE = c
        break
countries = dff[dff['Date'].eq(LAST_DATE) & dff['Confirmed Cases'].ge(MIN_CASES) & 
        dff['Country'].ne('China')
       ].sort_values(by='Confirmed Cases', ascending=False)
countries = countries['Country'].values



    











In [5]:



    


#hide
SINCE_CASES_NUM = 10
COL_X = f'Days since {SINCE_CASES_NUM}th death'
dff2 = dff[dff['Country'].isin(countries)].copy()
days_since = (dff2.assign(F=dff2['Confirmed Cases'].ge(SINCE_CASES_NUM))
              .set_index('Date')
              .groupby('Country')['F'].transform('idxmax'))
dff2[COL_X] = (dff2['Date'] - days_since.values).dt.days.values
dff2 = dff2[dff2[COL_X].ge(0)]



    











In [6]:



    


#hide
def get_country_colors(x):
    mapping = {
        'Italy': 'black',
        'Iran': '#A1BA59',
        'South Korea': '#E45756',
        'Spain': '#F58518',
        'Germany': '#9D755D',
        'France': '#F58518',
        'US': '#2495D3',
        'Switzerland': '#9D755D',
        'Norway': '#C1B7AD',
        'United Kingdom': '#2495D3',
        'Netherlands': '#C1B7AD',
        'Sweden': '#C1B7AD',
        'Belgium': '#C1B7AD',
        'Denmark': '#C1B7AD',
        'Austria': '#C1B7AD',
        'Japan': '#9467bd'}
    return mapping.get(x, '#C1B7AD')



    











In [7]:



    


#hide_input
baseline_countries = ['Italy', 'South Korea', 'Japan']
max_date = dff2['Date'].max()
color_domain = list(dff2['Country'].unique())
color_range = list(map(get_country_colors, color_domain))

def make_since_chart(highlight_countries=[], baseline_countries=baseline_countries):
    selection = alt.selection_multi(fields=['Country'], bind='legend', 
                                    init=[{'Country': x} for x in highlight_countries + baseline_countries])

    base = alt.Chart(dff2, width=550).encode(
        x=f'{COL_X}:Q',
        y=alt.Y('Confirmed Cases:Q', scale=alt.Scale(type='log'), axis=alt.Axis(title='Cumulative Deaths')),
        color=alt.Color(
            'Country:N',
            scale=alt.Scale(domain=color_domain, range=color_range),
            legend=alt.Legend(columns=len(color_domain)//18+1, symbolLimit=len(color_domain))),
        tooltip=list(dff2),
        opacity=alt.condition(selection, alt.value(1), alt.value(0.05))
    )
    max_day = dff2[COL_X].max()
    ref = pd.DataFrame([[x, SINCE_CASES_NUM*1.33**x] for x in range(max_day+1)], columns=[COL_X, 'Confirmed Cases'])
    base_ref = alt.Chart(ref).encode(x=f'{COL_X}:Q', y='Confirmed Cases:Q')
    return (
        base_ref.mark_line(color='black', opacity=.5, strokeDash=[3,3]) +
        base_ref.transform_filter(
            alt.datum[COL_X] >= max_day
        ).mark_text(dy=-6, align='right', fontSize=10, text='33% Daily Growth') +
        base.mark_line(point=True).add_selection(selection) + 
        base.transform_filter(
            alt.datum['Date'] >= int(max_date.timestamp() * 1000)
        ).mark_text(dy=-8, align='right', fontWeight='bold').encode(text='Country:N')
    ).properties(
        title=f"Compare {', '.join(highlight_countries)} death trajectory with {', '.join(baseline_countries)}"
    )

Learning from Italy, South Korea & Japan

Italy, South Korea & Japan are three countries which show different growth rates and how it evolved over time.

South Korea and Japan have lower growth rate since thier 10 deaths. Italy has been grow close to 33% till 3 weeks since it's 10 deaths.

Where does your Country stand today?

Click (Shift+ for multiple) on Countries legend to filter the visualization.





In [8]:



    


#hide_input
HTML(f'<small class="float-right">Last Updated on {pd.to_datetime(LAST_DATE).strftime("%B, %d %Y")}</small>')



    


















    
Out[8]:






Last Updated on March, 23 2020
















In [9]:



    


#hide_input
chart = make_since_chart()
chart



    


















    
Out[9]:


























In [10]:



    


#hide_input
chart2 = make_since_chart(['Spain', 'Germany'])
chart2



    


















    
Out[10]:


























In [11]:



    


#hide_input
chart3 = make_since_chart(['US', 'France'])
chart3



    


















    
Out[11]:


























In [12]:



    


#hide_input
chart4 = make_since_chart(['Germany', 'United Kingdom'])
chart4



    


















    
Out[12]:

Select a country from the drop down list below to toggle the visualization.





In [13]:



    


#hide_input
base = alt.Chart(dff2, width=600).encode(
    x=f'{COL_X}:Q',
    y=alt.Y('Confirmed Cases:Q', scale=alt.Scale(type='log'), axis=alt.Axis(title='Cumulative Deaths')),
    color=alt.Color('Country:N', scale=alt.Scale(domain=color_domain, range=color_range), legend=None),
    tooltip=['Country', 'Date', 'Confirmed Cases', COL_X]
)

country_selection = alt.selection_single(
    name='Select', fields=['Country'], 
    bind=alt.binding_select(options=list(sorted(set(countries) - set(baseline_countries)))),
    init={'Country': 'US'})

date_filter = alt.datum['Date'] >= int(max_date.timestamp() * 1000)
base2 = base.transform_filter(alt.FieldOneOfPredicate('Country', baseline_countries))
base3 = base.transform_filter(country_selection)
base4 = base3.transform_filter(date_filter)

max_day = dff2[COL_X].max()
ref = pd.DataFrame([[x, SINCE_CASES_NUM*1.33**x] for x in range(max_day+1)], columns=[COL_X, 'Confirmed Cases'])
base_ref = alt.Chart(ref).encode(x=f'{COL_X}:Q', y='Confirmed Cases:Q')
base_ref_f = base_ref.transform_filter(alt.datum[COL_X] >= max_day)

chart5 = (
 base_ref.mark_line(color='black', opacity=.5, strokeDash=[3,3]) + 
 base_ref_f.mark_text(dy=-6, align='right', fontSize=10, text='33% Daily Growth') + 
 base2.mark_line(point=True, tooltip=True) +
 base3.mark_line(point={'size':50}, tooltip=True) +
 base2.transform_filter(date_filter).mark_text(dy=-8, align='right').encode(text='Country:N') +
 base4.mark_text(dx=8, align='left', fontWeight='bold').encode(text='Country:N') +
 base4.mark_text(dx=8, dy=12, align='left', fontWeight='bold').encode(text='Confirmed Cases:Q')
).add_selection(country_selection).properties(
    title=f"Country's death trajectory compared to {', '.join(baseline_countries)}"
)
chart5



    


















    
Out[13]:

Interactive by Pratap Vardhan1

  1. Source: "2019 Novel Coronavirus COVID-19 (2019-nCoV) Data Repository by Johns Hopkins CSSE" GitHub repository. Link to original notebook.↩

Content source: pratapvardhan/iPyNotebooks

Similar notebooks:

notebook.community | gallery | about