%matplotlib inline

import os
import pandas as pd

df_public = pd.read_csv(os.path.expanduser('~/data/kaggle.com/stackoverflow/stack-overflow-2018-developer-survey/survey_results_public.csv'), usecols=['Country'], dtype=str)
df_countries = pd.DataFrame(df_public.Country.value_counts())
df_countries.head(5)

from iso3166 import countries

country_index = {name: countries.get(name).alpha3 for name in df_countries.index if name in countries}
set(df_countries.index) - set(country_index)

{'Bolivia',
 'Cape Verde',
 'Congo, Republic of the...',
 'Czech Republic',
 "Democratic People's Republic of Korea",
 'Democratic Republic of the Congo',
 'Hong Kong (S.A.R.)',
 'Iran, Islamic Republic of...',
 'Libyan Arab Jamahiriya',
 'Micronesia, Federated States of...',
 'North Korea',
 'Other Country (Not Listed Above)',
 'Republic of Korea',
 'Republic of Moldova',
 'South Korea',
 'The former Yugoslav Republic of Macedonia',
 'United Kingdom',
 'United Republic of Tanzania',
 'Venezuela, Bolivarian Republic of...'}

from collections import Counter

country_index.update({
 'Bolivia': 'BOL',
 'Cape Verde': 'CPV',
 'Congo, Republic of the...': 'COG',
 'Czech Republic': 'CZE',
 "Democratic People's Republic of Korea": 'PRK',
 'Democratic Republic of the Congo': 'COD',
 'Hong Kong (S.A.R.)': 'HKG',
 'Iran, Islamic Republic of...': 'IRN',
 'Libyan Arab Jamahiriya': 'LBY',
 'Micronesia, Federated States of...': 'FSM',
 'North Korea': 'PRK',
 'Republic of Korea': 'KOR',
 'Republic of Moldova': 'MDA',
 'South Korea': 'KOR',
 'The former Yugoslav Republic of Macedonia': 'MKD',
 'United Kingdom': 'GBR',
 'United Republic of Tanzania': 'TZA',
 'Venezuela, Bolivarian Republic of...': 'VEN'
})

pd.Series(country_index).value_counts().head()

PRK 2
KOR 2
ISL 1
HRV 1
ECU 1
dtype: int64

df_countries['iso'] = df_countries.index.map(lambda x: country_index.get(x))
iso_index = df_countries.groupby('iso').sum()
iso_index.sort_values('Country', ascending=False).head()

import geopandas as gpd

world = gpd.read_file(gpd.datasets.get_path('naturalearth_lowres')).to_crs('+proj=robin')
world = world[world.name != 'Antarctica']

world['respondents'] = world['iso_a3'].apply(lambda x: int(iso_index.loc[x]) if x in iso_index.index else None)
world['respondent_ratio'] = world['respondents'] / world['pop_est'] * 1_000_000
world.sort_values('respondent_ratio', ascending=False).head(10)

known = world.dropna(subset=['respondent_ratio'])
unknown = world[world['respondent_ratio'].isna()]

ax = known.plot(column='respondent_ratio', cmap='viridis_r', figsize=(20, 12), scheme='fisher_jenks', k=7, legend=True, edgecolor='#aaaaaa')
unknown.plot(ax=ax, color='#ffffff', hatch='//', edgecolor='#aaaaaa')

ax.set_title('Stack Overflow Developer Survey 2018 Respondents per 1 Million People', fontdict={'fontsize': 20}, loc='left')
descripton = '''
Survey data: kaggle.com/stackoverflow/stack-overflow-2018-developer-survey • Population estimates: naturalearthdata.com •
Source code: kaggle.com/ramirogomez/stack-overflow-survey-2018-respondents-world-map • Author: Ramiro Gómez - ramiro.org'''.strip()
ax.annotate(descripton, xy=(0.065, 0.12), size=12, xycoords='figure fraction')

ax.set_axis_off()
legend = ax.get_legend()
legend.set_bbox_to_anchor((.11, .4))
legend.prop.set_size(12)

%load_ext signature
%signature