The world's nuclear reactors over time

import pandas as pd, numpy as np
import matplotlib.pyplot as plt
import seaborn as sbs
Series, DataFrame = pd.Series, pd.DataFrame

# Make a dictionary of country names by code
country_names = {'KR': 'South Korea', 'GB': 'UK', 'RU': 'Russia',
    'CN': 'China', 'FR': 'France', 'JP': 'Japan', 'IN': 'India',
    'CA': 'Canada', 'UA': 'Ukraine'}

# CSV file column names.
names = ['type', 'status', 'location', 'refpower', 'grosscap',
         'dateconnected', 'country_code', 'dateshutdown', 'datesuspended',
         'datereconnected', 'id', 'name'
        ]
# Read in data file and convert dates to the correct type.
df = pd.read_csv('reactors.csv', names=names, header=None)
df['dateconnected'] = pd.to_datetime(df['dateconnected'])
df['dateshutdown'] = pd.to_datetime(df['dateshutdown'])
country_codes = df.country_code.unique()

reactors = DataFrame(0,index=range(1953, 2018), columns=country_codes)
for c_code in country_codes:
    c_df = df[df['country_code']==c_code][['dateconnected','dateshutdown']]
    reactors['on'] = c_df.dateconnected.dt.year.value_counts().sort_index()
    reactors['off'] = c_df.dateshutdown.dt.year.value_counts().sort_index()
    reactors['on'] = reactors['on'].fillna(0)
    reactors['off'] = reactors['off'].fillna(0)
    reactors[c_code] = (reactors['on'] - reactors['off']).cumsum()
reactors = reactors.drop(columns=['off', 'on'])

# We will label the nmost countries with the most nuclear reactors.
nmost = 10
most_nuclear_countries = reactors.loc[2017].sort_values()[-nmost:].index
least_nuclear_countries = reactors.columns ^ most_nuclear_countries
# Add the a column for the sum of the other countries' reactors and drop their
# individual columns.
reactors['Other'] = reactors[least_nuclear_countries].sum(axis=1)
plt_reactors = reactors.drop(columns=least_nuclear_countries)
plt_reactors = reactors[most_nuclear_countries]
plt_reactors['Other'] = reactors[least_nuclear_countries].sum(axis=1)

fig, ax = plt.subplots()
ax.stackplot(plt_reactors.index,
             *[plt_reactors.loc[:,cc] for cc in plt_reactors.columns],
             alpha=0.7, labels=plt_reactors.columns)

reactors_2017 = plt_reactors.loc[2017].cumsum()
# Find out where to locate the country label and annotate the graph
y = reactors_2017.rolling(2).mean().fillna(0)
for i,cc in enumerate(plt_reactors.columns):
    try:
        country_name = country_names[cc]
    except KeyError:
        country_name = cc
    ax.annotate(country_name, xy=(2018, y[i]))

ax.set_xlim(1954, 2030)
ax.set_xlabel('Year')
ax.set_ylabel('Number of reactors')
plt.savefig('reactors_by_country.png')
plt.show()