The Sunday tip #2: Measuring Python code performance with the timeit module

by Davide Mastromatteo
4 minute read Hi guys, this week's tip is about measuring the performance of your Python code. Maybe you left your cubicle last Friday with some piece of code that wasn't performing well? Well, today you will learn how to measure it! :)

The timeit module

We all know that Python comes with batteries included because in its standard library there are a lot of tools that you can easily use out of the box. One of theese tools is the timeit module.

The timeit module allows you to measure how much time is taken for a specific python piece of code. In short, it takes your piece of code, runs it a million times, and then returns you the execution time of the run.

Ok, let's start with an example. You have been asked to write a small function that takes a list of strings and returns a single string that is the concatenation of all the strings in the list.

Here is your piece of code:

1def concatenate(list: [str]):
2    result = ""
3    for string in list:
4        result = result + "," + string
6    return result

It was easy, wasn't it?

Now we need to test it… let's write some code to test it.

 1def concatenate(list: [str]):
 2    result = ""
 3    for string in list:
 4        result = result + "," + string
 6    return result
 8if __name__ == '__main__':
 9    import random
10    import string
12    my_list = []
13    for _ in range(100):
14        my_string = ""
15        for _ in range(10):
16            my_string = my_string + random.choice(string.ascii_uppercase) 
17        my_list.append(my_string)
19    result = concatenate(my_list)
20    print(result)

Ok, now we have added some code to create a random list of 100 ten-chars-strings and to call our concatenate() function. Running the program we can see that the program works… but how fast is it?

Let's find it out with the timeit() function of the timeit module!

The signature of the function we will use is the following:

1timeit.timeit(stmt='pass', setup='pass', timer=<default timer>, number=1000000, globals=None)

So in our code we can import the timeit module and use this function, where:

  • stmt is the statement to be tested, written as a string

  • setup is an optional string that you can use to setup the environment before starting

  • timer is an optional parameter to specify the timer we want to use (by default it is the time.perf_counter() timer)

  • number is the number of times the code has to be executed, by default is a million times

  • globals is an optional parameter that is useful if you want to specify a namespace where to execute your code

Let's try it:

 1def concatenate(list: [str]):
 2    result = ""
 3    for string in list:
 4        result = result + "," + string
 6    return result
 8if __name__ == '__main__':
 9    import random
10    import string
11    import timeit
13    my_list = []
14    for _ in range(100):
15        my_string = ""
16        for _ in range(10):
17            my_string = my_string + random.choice(string.ascii_uppercase) 
18        my_list.append(my_string)
20    print(timeit.timeit("concatenate(my_list)", globals=globals()))

Please note that in our example it was important to specify the globals parameter to specify in which namespace the module could have found the function we wrote. Running the example on my Intel-based MacBook pro what I got is:

$ python

Ok, let's face it, this code sucks guys…

Please note that during the execution, the timeit() function disables the garbage collector to have the single measurements more comparable. This is usually ok but sometimes it can be useful to measure it because garbage collections can be an important aspect to consider talking about performance. In this case, consider that you can reenable the garbage collection just by adding 'gc.enable()' to the setup parameter like this:

However, as you may know, there's a better way to concatenate strings in Python and it's done by using the .join() method of a string object. So in our case, our code could be written like this:

 1def concatenate(list: [str]):
 2    return ",".join(list)
 4if __name__ == '__main__':
 5    import random
 6    import string
 7    import timeit
 9    my_list = [("".join(random.choice(string.ascii_uppercase) for _ in range(10)) for _ in range(100)]
11    print(timeit.timeit("concatenate(my_list)", globals=globals()))

Yes, in the previous code we had reinvented the wheel… and by the way: if you are asking how I created the my_list list in this last example, check out my article about list comprehension.

Now, let's run this code and see what we get:

$ python

Ok, we have optimized our code by 95%, let's call it a day! ;)

So, what have we learned?

  • to measure a Python piece of code the best method is to use the timeit module of the standard library

  • optimizing your code is super important

  • the standard library is your friend :)

The timeit module has a lot of other features that we haven't discussed in this article, to find out more, refer to the standard library official documentation.

Happy Sunday!


Did you find this article helpful?