| Index: trunk/tools/wsor/contribution_inequality/ginichart.py |
| — | — | @@ -0,0 +1,104 @@ |
| | 2 | +#!/usr/bin/python |
| | 3 | +''' computes gini coefficient of contribution to namespace per year ''' |
| | 4 | + |
| | 5 | +import os |
| | 6 | +import sys |
| | 7 | +import csv |
| | 8 | + |
| | 9 | +import numpy as np |
| | 10 | +import matplotlib.pyplot as pp |
| | 11 | + |
| | 12 | +from itertools import groupby |
| | 13 | +from contextlib import closing |
| | 14 | +from argparse import ArgumentParser |
| | 15 | +from matplotlib.font_manager import FontProperties |
| | 16 | + |
| | 17 | +parser = ArgumentParser(description=__doc__) |
| | 18 | +parser.add_argument('data_path', metavar='data') |
| | 19 | +parser.add_argument('-T', '--title') |
| | 20 | + |
| | 21 | +colors = 'bgrcmykw' |
| | 22 | +styles = ['-', '--', '-.', ':'] |
| | 23 | +markers = 'ov^<>1234' |
| | 24 | + |
| | 25 | +def gini(x): |
| | 26 | + ''' |
| | 27 | + Computes an estimator of the Gini coefficient from an array x |
| | 28 | + Parameters |
| | 29 | + ---------- |
| | 30 | + x - a flat array of observations |
| | 31 | + |
| | 32 | + References |
| | 33 | + ---------- |
| | 34 | + http://mathworld.wolfram.com/GiniCoefficient.html |
| | 35 | + ''' |
| | 36 | + x.sort() # sorts in non-decreasing order |
| | 37 | + n = float(len(x)) |
| | 38 | + i = np.arange(len(x)) + 1 |
| | 39 | + m = np.mean(x) |
| | 40 | + return np.sum((2 * i - n - 1) * x) / ( n ** 2 * m) * (n / (n - 1)) |
| | 41 | + |
| | 42 | +def igini(flatiter): |
| | 43 | + ''' |
| | 44 | + Computes an estimator of the Gini coefficient from a sorted iterator on a |
| | 45 | + flat sample of observations |
| | 46 | + |
| | 47 | + Parameters |
| | 48 | + ---------- |
| | 49 | + flatiter - an iterator over observations, sorted in non-decreasing order |
| | 50 | + |
| | 51 | + References |
| | 52 | + ---------- |
| | 53 | + http://en.wikipedia.org/wiki/Gini_coefficient |
| | 54 | + http://mathworld.wolfram.com/GiniCoefficient.html |
| | 55 | + ''' |
| | 56 | + den = 0.0 |
| | 57 | + num = 0.0 |
| | 58 | + for i, y in enumerate(flatiter): |
| | 59 | + num += (i + 1) * y |
| | 60 | + den += y |
| | 61 | + n = i + 1 |
| | 62 | + return 1 - (2.0 / (n - 1)) * (n - num / den) * (n / (n - 1)) |
| | 63 | + |
| | 64 | +if __name__ == '__main__': |
| | 65 | + |
| | 66 | + ns = parser.parse_args() |
| | 67 | + |
| | 68 | + g = [] |
| | 69 | + |
| | 70 | + with closing(open(ns.data_path)) as f: |
| | 71 | + reader = csv.DictReader(f, delimiter='\t', quoting=csv.QUOTE_NONE) |
| | 72 | + groupfunc = lambda row : map(int, (row['namespace'], row['year'])) |
| | 73 | + for key, subiter in groupby(reader, groupfunc): |
| | 74 | + flatiter = ( float(row['total_contributions']) for row in subiter ) |
| | 75 | + try: |
| | 76 | + g.append((tuple(key) + (igini(flatiter),))) |
| | 77 | + except ZeroDivisionError: # due to passing an empty iterator to igini |
| | 78 | + g.append((tuple(key) + (np.nan,))) |
| | 79 | + |
| | 80 | + figure = pp.figure(figsize=(8,4)) |
| | 81 | + ax = figure.add_axes(pp.axes([.1,.1,.8,.8], axisbg='whitesmoke')) |
| | 82 | + i = 0 |
| | 83 | + M = len(markers) |
| | 84 | + C = len(colors) |
| | 85 | + S = len(styles) |
| | 86 | + |
| | 87 | + for key, subiter in groupby(g, lambda k : k[0]): |
| | 88 | + data = np.asarray([ (y,g) for n, y, g in subiter ]) |
| | 89 | + label = 'NS %d' % key |
| | 90 | + ax.plot(data.T[0], data.T[1], label=label, marker=markers[i % M], |
| | 91 | + color=colors[i % C], linestyle=styles[i % S]) |
| | 92 | + i += 1 |
| | 93 | + |
| | 94 | + pp.ylabel('Gini coefficient') |
| | 95 | + pp.legend(loc='best', prop=FontProperties(size='small')) |
| | 96 | + pp.ylim(0,1) |
| | 97 | + pp.draw() |
| | 98 | + if ns.title: |
| | 99 | + pp.title(ns.title) |
| | 100 | + figure_path = 'gini_' + ns.title.replace(' ', '_') + '.pdf' |
| | 101 | + else: |
| | 102 | + figure_path = 'gini_' + os.path.splitext(ns.data_path)[0] + '.pdf' |
| | 103 | + pp.savefig(figure_path, fmt='pdf') |
| | 104 | + print 'output saved to %s' % figure_path |
| | 105 | + pp.show() |
| Property changes on: trunk/tools/wsor/contribution_inequality/ginichart.py |
| ___________________________________________________________________ |
| Added: svn:executable |
| 1 | 106 | + * |