A test comparing defaultdict and groupby for grouping

gistfile1.py

from timeit import Timer
from operator import attrgetter
from random import randint

def timeit( stmt, setup ):
	t = Timer(stmt=stmt, setup=setup)
	print "%.2f usec/pass" % (100000 * t.timeit(number=10000)/10000)

class Person(object):
	def __init__(self, age):
		self.age = age
	def __repr__(self):
		return str(self.age)

persons = [Person(age) for age in (randint(10,90) for _ in range(1000))]

get_age = attrgetter('age')

setup0 = """
from collections import defaultdict
from __main__ import persons
"""

setup1 = """
from itertools import groupby
from __main__ import persons, get_age
"""

stmt0="""
persons_by_age = defaultdict(list)
for person in persons:
	persons_by_age[person.age].append(person)
"""

stmt1="""
persons_by_age = {k: list(g) for k, g in groupby(sorted(persons, key=get_age), get_age)}
"""

timeit(stmt=stmt0, setup=setup0)
timeit(stmt=stmt1, setup=setup1)

On my CPU the output is

29.74 usec/pass
103.87 usec/pass