population plot

plot

import numpy as np
import matplotlib.pyplot as plt

pops = np.array([39538223,
29145505,
21538187,
20201249,
13011844,
12812508,
11799448,
10711908,
10439388,
10077331,
9288994,
8631393,
7705281,
7151502,
7029917,
6910840,
6785528,
6177224,
6154913,
5893718,
5773714,
5706494,
5118425,
5024279,
4657757,
4505836,
4237256,
3959353,
3605944,
3271616,
3190369,
3104614,
3011524,
2961279,
2937880,
2117522,
1961504,
1839106,
1793716,
1455271,
1377529,
1362359,
1097379,
1084225,
989948,
886667,
779094,
733391,
643077,
576851])

cumulative_pop = np.concatenate([np.array([0]), np.cumsum(np.sort(pops))])
total = cumulative_pop.max()



fig, ax = plt.subplots(1, figsize=(5, 5))
for ii in range(50):
    start_frac = 100*cumulative_pop[ii] / total
    end_frac = 100*cumulative_pop[ii+1] / total
    start_sen = ii*2
    end_sen = (ii+1)*2
    x = np.linspace(start_frac, end_frac, 10)
    y = np.linspace(start_sen, end_sen, 10)
    ax.plot(x, y, c='k')
    ax.scatter(end_frac, end_sen, marker='.', c='k', s=10)
    if (ii % 10) == 0 and (ii > 0):
        ax.annotate(f'{int(np.around(end_frac))}%, {end_sen}',
                    (end_frac, end_sen),
                    (end_frac + 20, end_sen - 20),
                    arrowprops={'arrowstyle': '->'})
ax.set_xlim(0, 100)
ax.set_ylim(0, 100)
ax.plot([0, 100], [0, 100], '--', c='k')
ax.set_xlabel('Percent of US population', fontsize=12)
ax.set_ylabel('Number of US senators', fontsize=12)
fig.tight_layout()
plt.savefig('pop_vs_senate.png', dpi=300)