http://www.mediawiki.org/wiki/Special:Code/MediaWiki/94957
Revision: 94957
Author: giovanni
Date: 2011-08-18 22:59:04 +0000 (Thu, 18 Aug 2011)
Log Message:
-----------
now topcontributors.py plots the set similarity of top users over the years
Modified Paths:
--------------
trunk/tools/wsor/contribution_inequality/topcontributors.py
Modified: trunk/tools/wsor/contribution_inequality/topcontributors.py
===================================================================
--- trunk/tools/wsor/contribution_inequality/topcontributors.py 2011-08-18
22:04:42 UTC (rev 94956)
+++ trunk/tools/wsor/contribution_inequality/topcontributors.py 2011-08-18
22:59:04 UTC (rev 94957)
@@ -8,25 +8,77 @@
from itertools import groupby
from contextlib import closing
-from argparse import ArgumentParser
+from argparse import ArgumentParser, FileType
from collections import deque
+from matplotlib.font_manager import FontProperties
+from datetime import date
+import numpy as np
+import matplotlib.pyplot as pp
+
parser = ArgumentParser(description=__doc__)
parser.add_argument('data_path', metavar='data')
-parser.add_argument('maxlen', metavar='number', type=int)
+parser.add_argument('output_file', metavar='output_file', type=FileType('w'))
+parser.add_argument('-t', '--top', dest='maxlen', type=int, default=100,
+ help='Top users to list. default: %(default)d', metavar='NUM')
+colors = 'bgrcmykw'
+styles = ['-', '--', '-.', ':']
+markers = 'ov^<>1234'
+
if __name__ == '__main__':
ns = parser.parse_args()
-
+ databyns = {}
+
with closing(open(ns.data_path)) as f:
reader = csv.DictReader(f, delimiter='\t', quoting=csv.QUOTE_NONE)
groupfunc = lambda row : (row['namespace'], row['year'])
for key, subiter in groupby(reader, groupfunc):
# smart way to keep only the tail
users = deque((row['user_id'] for row in subiter ),
maxlen=ns.maxlen)
- print '\t'.join(key + tuple(users))
- sys.stdout.flush()
+ print >> ns.output_file, '\t'.join(key + tuple(users))
+ ns.output_file.flush()
+ NS, year = map(int, key)
+ try:
+ databyns[NS].append((year, set(users)))
+ except KeyError:
+ databyns[NS] = [ (year, set(users)) ]
+
+ figure = pp.figure(figsize=(8,4))
+ ax = figure.add_axes(pp.axes([.1,.1,.8,.8], axisbg='whitesmoke'))
+ i = 0
+ M = len(markers)
+ C = len(colors)
+ S = len(styles)
+
+ for key in databyns:
+ years, users = zip(*databyns[key])
+ years = [ date(year, 1, 1) for year in years ]
+ I = np.asfarray(map(len, map(set.intersection, users[1:], users[:-1])))
+ U = np.asfarray(map(len, map(set.union, users[1:], users[:-1])))
+ label = 'NS %s' % key
+
+ ax.plot(years[1:], I / U, label=label, marker=markers[i % M],
+ color=colors[i % C], linestyle=styles[i % S])
+ i += 1
+
+ pp.ylim(0,1)
+ pp.ylabel('similarity')
+ pp.title('Top %d contributors' % ns.maxlen)
+ pp.legend(loc='best', prop=FontProperties(size='small'))
+ pp.draw()
+
+ if not ns.output_file.isatty():
+ figure_path = os.path.splitext(ns.output_file.name)[0] + '.pdf'
+ pp.savefig(figure_path, fmt='pdf')
+ print 'figure saved to %s' % figure_path
+ print 'output saved to %s' % ns.output_file.name
+
+ pp.show()
+ if not ns.output_file.isatty():
+ ns.output_file.close()
+
_______________________________________________
MediaWiki-CVS mailing list
MediaWiki-CVS@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/mediawiki-cvs
