projects.mako.cc - matplotlib-cdsw/blob - fractionplot.py

   1 """Plot the fraction of articles that are female, by birth year and first edit date."""
   2 import wikibios
   3 from matplotlib import pyplot
   4 from operator import itemgetter
   5
   6 figure = pyplot.figure()
   7
   8 rows_by_birth_year = sorted(wikibios.rows, key=itemgetter('birth_year'))
   9 birth_year_medians = []
  10 fraction_female_by_birth_year = []
  11 N = 1000
  12 i = 0
  13 while i + N <= len(rows_by_birth_year):
  14         chunk = rows_by_birth_year[i:i+N]
  15         i = i + N
  16
  17         birth_year_medians.append(chunk[N / 2]['birth_year'])
  18
  19         count_female = 0.0
  20         for row in chunk:
  21                 if row['gender'] == 'female':
  22                         count_female = count_female + 1
  23         fraction_female = count_female / N
  24         fraction_female_by_birth_year.append(fraction_female)
  25
  26 axes1 = figure.add_subplot(2, 1, 1)
  27 axes1.plot(birth_year_medians, fraction_female_by_birth_year)
  28 axes1.set_xlabel('Birth Year')
  29 axes1.set_ylabel('Fraction Female')
  30
  31 rows_by_firstedit = sorted(wikibios.rows, key=itemgetter('firstedit'))
  32 fraction_female_by_firstedit = []
  33 firstedit_medians = []
  34 i = 0
  35 while i + N <= len(rows_by_birth_year):
  36         chunk = rows_by_firstedit[i:i+N]
  37         i = i + N
  38
  39         firstedit_medians.append(chunk[N / 2]['firstedit'])
  40
  41         count_female = 0.0
  42         for row in chunk:
  43                 if row['gender'] == 'female':
  44                         count_female = count_female + 1
  45         fraction_female = count_female / N
  46         fraction_female_by_firstedit.append(fraction_female)
  47
  48 axes2 = figure.add_subplot(2, 1, 2)
  49 axes2.plot(firstedit_medians, fraction_female_by_firstedit)
  50 axes2.set_xlabel('Article Year')
  51 axes2.set_ylabel('Fraction Female')
  52
  53 figure.savefig('fractions.pdf')