benfords-lots.py

import json
from collections import Counter, defaultdict
import requests
import sys
import matplotlib.pyplot as plt


def loads_and_parse(nums):
  parse_f = lambda s : int(s.replace(',', ''))
  return list(map(parse_f, json.loads(nums)))


def get_data(state_name):
  ret = defaultdict(list)
  counties = requests.get(f"https://static01.nyt.com/elections-assets/2020/data/api/2020-11-03/state-page/{state_name}.json").json()['data']['races'][0]['counties']
  for county in counties:
    for cand, votes in county['results'].items():
      ret[cand].append(votes)
  return ret

def first_digits_kvs(kvs):
  k, vs = kvs
  # return (k, Counter(list(map(lambda v: int(str(v)[0]), vs))))
  return (k, list(map(lambda v: int(str(v)[0]), vs)))

# from https://realpython.com/python-histograms/
def ascii_histogram(seq) -> None:
    """A horizontal frequency-table/histogram plot."""
    for k in sorted(seq):
        print('{0:5d} {1}'.format(k, '+' * seq[k]))
        
def heading(h):
  print(f"\n\n{h} histogram")
 
def histogram_state(state, counts, only_two=True):
  if only_two:
    counts = dict(bidenj=counts['bidenj'], trumpd=counts['trumpd'])
  plt.figure(figsize=([16,9]))
  leg = list(counts.keys())
  print(leg)
  zipped_cs = list(map(list, counts.values()))
  print(zipped_cs)
  n, bins, patches = plt.hist(zipped_cs, 9, density=True, histtype='bar', label=leg, color=['blue', 'red'])
  plt.legend(prop=dict(size=10))
  # plt.grid(axis='y', alpha=0.75)
  plt.xlabel('leading_digit')
  plt.ylabel('frequency')
  plt.title(f'benford histogram for {state}')
  plt.savefig(f"b-out-{state}.png")

   
def print_graph(h, seq):
  heading(h)
  ascii_histogram(seq)

for state in ['illinois', 'nevada', 'wisconsin', 'washington', 'new-york', 'florida', 'arizona', 'georgia', 'wyoming', 'california', 'pennsylvania', 'ohio']:
  try:
    print(f"\n\n  ## {state} ##\n")
    cand_votes = get_data(state)
    pairs = dict(map(first_digits_kvs, cand_votes.items()))
    histogram_state(state, pairs)
#    for k, v in pairs.items():
#      print_graph(k, v)
  except Exception as e:
    print(f'exception during state: {state}')
    raise e