How do I calculate the average running time of a function?

Question

To calculate the execution time of a function I am using

inicio=time.time()
# código 
fim=time.time()

How do I calculate the time repeatedly and then calculate the average of that time?

Answer 1

You can place the snippet of code you are measuring execution within a loop:

import time

REPETICOES = 10
tempos = []
for vezes in range(0, REPETICOES):
    inicio = time.time()
    # ... o código a medir entra aqui ... #
    fim = time.time()
    tempos.append(fim - inicio)

media = sum(tempos) / REPETICOES

Or use the module timeit:

import timeit

REPETICOES = 10

# o seu código fica dentro de uma string mesmo
meu_codigo = """
j = 0
for i in range(0,999):
    j += i * j
"""

tempo = timeit.timeit(stmt=meu_codigo, number=REPETICOES)
print(tempo / REPETICOES)

But you can see that in both, whether "manually" or through the library, the implementations are quite similar.

Answer 2

An interesting way that Python lets you do it is to create a decorator that manages it for you:

def time_statistics(func):
    def wrapper(*args, **kwargs):
        start = time.time()
        result = func(*args, **kwargs)
        end = time.time()

        if not hasattr(wrapper, 'times'):
            wrapper.times = []

        wrapper.times.append(end - start)
        return result

    return wrapper

So in the function you want to analyze, just do:

@time_statistics
def foo(a, b):
    return a + b

Make calls to test time:

for _ in range(100):
    foo(1, 2)

Then analyze the results:

print('Tempo mínimo [s]: ', min(foo.times))
print('Tempo máximo [s]: ', max(foo.times))
print('Média de tempo [s]: ', sum(foo.times) / len(foo.times))

See working on Repl.it.

Tip: I used a simple list to manage the times, but you can easily abstract it into a more complex structure that already manages all the logic you need, whether it’s calculating minimums and maximums, calculating the mean, standard deviation, variance, or generating the graph.