Задача о сравнении подходов

find_max.py

import random
from collections import defaultdict
from functools import reduce
from timeit import timeit
from itertools import groupby
from operator import itemgetter
from statistics import mean

AMAX = 65

l = [{'age': random.randint(18, AMAX), 'rate': random.randint(1, 1000)}
     for x in range(100000)]

# ========================


def max_for(l):
    totals = [0 for i in range(AMAX + 1)]
    counters = [0 for i in range(AMAX + 1)]

    for p in l:
        totals[p['age']] += p['rate']
        counters[p['age']] += 1

    averages = []

    for i, v in enumerate(counters):
        if counters[i] > 0:
            averages.append(totals[i] / counters[i])
    return max(averages)
# ========================

# ========================


def max_map(l):
    def fn(a, b):
        a[b['age']].append(b['rate'])
        return a
    return max(
        map(
            lambda x: sum(x) / len(x),
            reduce(fn, [defaultdict(list)] + l).values()
        )
    )
# ========================


def max_fun(l):
    return max([mean(i['rate'] for i in g)
                for k, g in groupby(l, key=itemgetter('age'))])


t1 = timeit('max_for(l)', 'from __main__ import max_for, l', number=1)
t2 = timeit('max_map(l)', 'from __main__ import max_map, l', number=1)
t3 = timeit('max_fun(l)', 'from __main__ import max_fun, l', number=1)
print(t1)
print(t2)
print(t3)

The results between max_fun(l) and max_for(l) | max_map(l) are not equal if we take l from code above. But everything is ok when test list is small. So, there is some logical bug in max_fun (or vice versa in max_for | max_map?).

l = [{'age': random.randint(18, 65), 'rate': random.randint(1, 1000)} for x in range(100000)]
print(max_for(l)) #>> 512.7681692732291
print(max_map(l)) #>> 512.7681692732291
print(max_fun(l)) #>> 1000

test = [{'age': 18, 'rate': 30}, {'age': 18, 'rate': 15}, {'age': 50, 'rate': 35}]
print(max_for(test)) #>> 35.0
print(max_map(test)) #>> 35.0
print(max_fun(test)) #>> 35