greenivlle.py

import pandas as pd
import matplotlib.pyplot as plt
data = pd.read_csv('data/Greenville_County_School_District_Spending.csv',
                   converters={'Amount': lambda x: float(x.replace('$', ''))},
                   parse_dates=9)
singleDigitCounts = data\
  .Amount[data.Amount >= 1]\
  .apply(lambda x: int(str(x)[:1]))\
  .value_counts()
  
def compareVisuallyToBenford(counts, plotTitle):
  """Compare digit counts visually with Benford's law.
  
  Will create a plot.
  
  Inputs
  ------
  
  counts: A pandas series of counts with digits as index.
  plotTitle: The plot's title.
  
  """
  total = counts.sum()
  indexSorted = np.array(sorted(counts.index))
  plt.plot(indexSorted, np.log10(1 + 1./indexSorted) * total, label='Benford\'s Law')
  plt.bar(counts.index.values - 0.4, counts, label='Actual Counts')
  plt.title(plotTitle)
  plt.legend()
  
if __name__ == '__main__':
  compareVisuallyToBenford(singleDigitCounts, 'First Digit Counts')