In [1]:

import numpy as np
import pandas as pd

from bokeh.charts import Histogram, Bar, BoxPlot
from bokeh.plotting import figure, show
from bokeh.io import output_notebook

Ebola outbreaks data before 2014¶

In [2]:

ebola_df = pd.read_csv("data/out/ebola_outbreaks_before_2014-geometry_fixed.csv", encoding="utf-8", index_col=False)

ebola_data = ebola_df.drop(ebola_df.columns[[0, 1, 2, 9]], axis=1)

Data set description¶

In [3]:

cols = list(ebola_data.columns)

for column in cols:
    print "column", (cols.index(column) + 1), ":", column

column 1 : country_code_iso_2_digits
column 2 : country_name
column 3 : duration_days
column 4 : ebola_subtype
column 5 : end_date
column 6 : end_datetime
column 7 : geometry
column 8 : geometry_geojson
column 9 : latitude
column 10 : longitude
column 11 : reported_number_of_deaths_among_cases
column 12 : reported_number_of_human_cases
column 13 : reported_of_deaths_among_cases
column 14 : start_date
column 15 : start_datetime
column 16 : year_s

=> locations info, reported cases, date/time

Number of outbreaks per countries¶

In [4]:

ebola_data.groupby(["country_name"])["country_name"].count().order()

Out[4]:

country_name
Côte d'Ivoire (Ivory Coast)         1
England                             1
Italy                               1
South Africa                        1
Russia                              2
Philippines                         3
Sudan (South Sudan)                 3
USA                                 3
Gabon                               4
Uganda                              5
Democratic Republic of the Congo    9
Name: country_name, dtype: int64

In [4]:

# Bokeh can't seem to handle the unicode for circumflex accent on the 'o' in "Côte d'Ivoire"
# I'm replacing it by a normal 'o'

countries_list = list(ebola_data["country_name"])

for i in range(len(countries_list)): 
    if countries_list[i] == u"C\xf4te d'Ivoire (Ivory Coast)":
       countries_list[i] = u"Cote d'Ivoire (Ivory Coast)"

In [5]:

data_nb = {
    'countries': countries_list,
    'ebola outbreaks': [1] * len(countries_list)
}

bar_nb = Bar(data_nb, values='ebola outbreaks', label='countries', agg='sum', color="#3B6849",
          title="Number of ebola outbreak(s) per country", plot_width=600, plot_height=500)


output_notebook()

show(bar_nb)

Out[5]:

<Bokeh Notebook handle for In[5]>

The average number of days for an outbreak per country¶

In [6]:

ebola_data.groupby(["country_name"])["duration_days"].mean().order()

Out[6]:

country_name
Uganda                              206.800000
Gabon                               219.750000
Democratic Republic of the Congo    235.444444
Côte d'Ivoire (Ivory Coast)         364.000000
USA                                 364.333333
Philippines                         364.666667
Sudan (South Sudan)                 364.666667
England                             365.000000
Italy                               365.000000
Russia                              365.000000
South Africa                        365.000000
Name: duration_days, dtype: float64

In [7]:

data_dur = {
    'countries': countries_list,
    'outbreaks duration': list(ebola_data["duration_days"])
}

bar_dur = Bar(data_dur, values='outbreaks duration', label='countries', agg='mean', color="#586996",
              title="Durations of ebola outbreaks (days) per country", plot_width=600, plot_height=500)

output_notebook()

show(bar_dur)

Out[7]:

<Bokeh Notebook handle for In[7]>

Ebola virus subtypes¶

In [9]:

ebola_data.groupby(["ebola_subtype"])["ebola_subtype"].count().order()

Out[9]:

ebola_subtype
Taï Forest virus     1
Bundibugyo virus     2
Reston virus         7
Sudan virus          8
Zaire virus         15
Name: ebola_subtype, dtype: int64

In [10]:

# Bokeh can't seem to handle the unicode for trema accent on the 'i' in Taï Forest virus
# I'm replacing it by a normal 'i'

eb_virus_types = list(ebola_data["ebola_subtype"])

for i in range(len(eb_virus_types)): 
    if eb_virus_types[i] == u"Ta\xef Forest virus":
       eb_virus_types[i] = u"Tai Forest virus"

In [11]:

data_vir = {
    'ebola virus subtypes': eb_virus_types,
    'number of ebola outbreaks': [1] * len(eb_virus_types)
}

bar_vir = Bar(data_vir, values='number of ebola outbreaks', label='ebola virus subtypes', agg='sum', color="#E2AE7A",
              title="Number of ebola outbreak(s) per virus subtype", plot_width=600, plot_height=400, bar_width=0.5)

output_notebook()

show(bar_vir)

Out[11]:

<Bokeh Notebook handle for In[11]>

In [12]:

ebola_data[["country_name", "ebola_subtype"]].sort(["country_name"])

Out[12]:

	country_name	ebola_subtype
30	Côte d'Ivoire (Ivory Coast)	Taï Forest virus
16	Democratic Republic of the Congo	Zaire virus
18	Democratic Republic of the Congo	Zaire virus
17	Democratic Republic of the Congo	Zaire virus
15	Democratic Republic of the Congo	Bundibugyo virus
5	Democratic Republic of the Congo	Zaire virus
6	Democratic Republic of the Congo	Zaire virus
12	Democratic Republic of the Congo	Zaire virus
8	Democratic Republic of the Congo	Zaire virus
10	Democratic Republic of the Congo	Zaire virus
32	England	Sudan virus
13	Gabon	Zaire virus
4	Gabon	Zaire virus
20	Gabon	Zaire virus
23	Gabon	Zaire virus
24	Italy	Reston virus
7	Philippines	Reston virus
14	Philippines	Reston virus
1	Philippines	Reston virus
0	Russia	Zaire virus
19	Russia	Zaire virus
27	South Africa	Zaire virus
31	Sudan (South Sudan)	Sudan virus
29	Sudan (South Sudan)	Sudan virus
28	Sudan (South Sudan)	Sudan virus
21	USA	Reston virus
2	USA	Reston virus
3	USA	Reston virus
26	Uganda	Sudan virus
22	Uganda	Bundibugyo virus
9	Uganda	Sudan virus
25	Uganda	Sudan virus
11	Uganda	Sudan virus

Duration of epidemy in days by virus subtype¶

In [13]:

ebola_data.groupby(["ebola_subtype"])["duration_days"].mean().order()

Out[13]:

ebola_subtype
Bundibugyo virus    121.500000
Zaire virus         260.733333
Sudan virus         304.000000
Taï Forest virus    364.000000
Reston virus        364.571429
Name: duration_days, dtype: float64

=> Bundibugyo virus seems to be correlated with shorter outbreaks

Human cases for DRC (country with the most cases)¶

In [14]:

df_drc = ebola_data[ebola_data.country_name == "Democratic Republic of the Congo"]

drc = df_drc[["ebola_subtype", "start_date", "end_date", "reported_number_of_human_cases", "reported_number_of_deaths_among_cases"]]

drc.sort(["start_date"])

Out[14]:

	ebola_subtype	start_date	end_date	reported_number_of_human_cases	reported_number_of_deaths_among_cases
5	Zaire virus	1976-01-01T00:00:00Z	1976-12-31T00:00:00Z	318	280
8	Zaire virus	1977-01-01T00:00:00Z	1977-12-31T00:00:00Z	1	1
18	Zaire virus	1995-01-01T00:00:00Z	1995-12-31T00:00:00Z	315	250
16	Zaire virus	2001-10-01T00:00:00Z	2002-03-31T00:00:00Z	57	43
17	Zaire virus	2002-12-01T00:00:00Z	2003-04-30T00:00:00Z	143	128
10	Zaire virus	2003-11-01T00:00:00Z	2003-12-31T00:00:00Z	35	29
6	Zaire virus	2007-01-01T00:00:00Z	2007-12-31T00:00:00Z	264	187
12	Zaire virus	2008-12-01T00:00:00Z	2009-02-28T00:00:00Z	32	15
15	Bundibugyo virus	2012-06-01T00:00:00Z	2012-11-30T00:00:00Z	36	13