Data on Policemen per Country and Population¶

IPython Notebook to retrieve, cleanup, display and plot data on number of policemen and their population ratio for over 100 countries from Wikipedia.

In [1]:

import requests
import io
import re

import pandas as pd
import geonamescache

from geonamescache import mappings

gc = geonamescache.GeonamesCache()
cnames = gc.get_countries_by_names()

url = 'http://wikitables.geeksta.net/dl/?url=https%3A%2F%2Fen.wikipedia.org%2Fwiki%2FList_of_countries_by_number_of_police_officers&idx=0'

re_num = re.compile(r'^[\d,.]+$')

def fix_num(x):
    if (isinstance(x, str) and re.search(re_num, x)):
        x = x.replace(',', '')
        if '.' in x:
            x = float(x)
        else:
            x = int(x)
    return x

Get and cleanup data

Convert number strings to floats or integers, remove trailing numbers (footnotes) from country names and replace country names with those used in geonamescache. Also add an iso3 column for rendering the d3 based map.

In [2]:

csv = requests.get(url).text
df = pd.read_csv(io.StringIO(csv))
df = df.applymap(fix_num)

num = ''.join([str(i) for i in range(10)])
df['Country'] = df['Country'].apply(lambda x: x.rstrip(num))
df['Country'] = df['Country'].apply(lambda x: mappings.country_names.get(x, x))
df['iso3'] = df['Country'].apply(lambda x: cnames[x]['iso3'])
df.set_index('Country', inplace=True)
df.head()

Out[2]:

	Size	Year	Police per 100,000 people	iso3
Country
Afghanistan	122000	2012	401	AFG
American Samoa	200	2012	720	ASM
Andorra	237	2012	278	AND
Antigua and Barbuda	600	2012	733	ATG
Argentina	205902	2000	558	ARG

Display and plot most and least policed countries

First set some common plotting properties.

In [3]:

footer = 'CC BY-SA 2014 Ramiro Gómez - ramiro.org • Data: en.wikipedia.org/wiki/List_of_countries_by_number_of_police_officers'
mpl.rcParams['font.size'] = 11
mpl.rcParams['font.family'] = 'Ubuntu'
mpl.rcParams['axes.color_cycle'] = 'a6cee3, 1f78b4, b2df8a, 33a02c, fb9a99, e31a1c, fdbf6f, ff7f00, cab2d6'

Show most policed countries

In [4]:

df.sort('Police per 100,000 people', inplace=True)
df.tail(10)

Out[4]:

	Size	Year	Police per 100,000 people	iso3
Country
Niue	16	2012	800	NIU
Grenada	900	2012	818	GRD
Montenegro	5250	2012	839	MNE
Bahamas	3000	2012	848	BHS
Saint Kitts and Nevis	450	2012	899	KNA
Saint Helena	69	2012	939	SHN
Brunei	4400	2012	1076	BRN
Monaco	500	2012	1374	MCO
Pitcairn	1	2012	1492	PCN
Vatican	130	2012	15625	VAT

Plot most policed countries

In [5]:

info = 'Number of policemen by 100,000 people.\n'

s = df['Police per 100,000 people'][-10:]
s.plot(kind='barh', figsize=(10, 6), title='Most Policed Countries\n', fontsize='large')

ax = plt.axes()
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
ax.spines['bottom'].set_visible(False)
ax.spines['left'].set_color((1, 1, 1))
ax.xaxis.set_label_text(info + footer)
ax.xaxis.set_ticklabels('')
ax.yaxis.set_label_text('')

for i, x in enumerate(s):
    ax.text(x + 100, i - .1, x, ha='left', fontsize='large')

plt.savefig('../static/img/graphs/most-policed-countries.png', bbox_inches='tight')

Show least policed countries

In [6]:

df.head(10)

Out[6]:

	Size	Year	Police per 100,000 people	iso3
Country
Mali	7000	2012	48	MLI
Niger	8700	2012	58	NER
Comoros	500	2012	66	COM
Togo	4000	2012	72	TGO
Iran	60000	2012	80	IRN
Kenya	35000	2012	81	KEN
French Polynesia	220	2012	82	PYF
Bangladesh	135000	2012	83	BGD
Ghana	23000	2012	94	GHA
Guinea	10000	2012	100	GIN

Plot least policed countries

In [7]:

info = 'Number of policemen by 100,000 people.\n'

s = df['Police per 100,000 people'][:10]
s.plot(kind='barh', figsize=(10, 6), title='Least Policed Countries\n', fontsize='large')

ax = plt.axes()
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
ax.spines['bottom'].set_visible(False)
ax.spines['left'].set_color((1, 1, 1))
ax.xaxis.set_label_text(info + footer)
ax.xaxis.set_ticklabels('')
ax.yaxis.set_label_text('')

for i, x in enumerate(s):
    ax.text(x + 1, i - .1, x, ha='left', fontsize='large')

plt.savefig('../static/img/graphs/least-policed-countries.png', bbox_inches='tight')

In [8]:

df.columns = ['Size', 'Year', 'Police per 100,000 People', 'iso3']
df.to_csv('../static/data/csv/police-countries.csv', encoding='utf-8', index=False)