Source: 📖 Effective Python item 70

Using Profile

The cProfile module offers functionality to measure the runtime performance of different objects. You simply make a Profile object, feed it the function you'l like to measure and then use Stats from the pstats module to print out statistics.

from cProfile import Profile
from pstats import Stats

profiler = Profile
profiler.runcall(test)		# measureing a 'test' function

stats = Stats(profiler)		# making a stats object of the profile's data
stats.print_stats()		# print statistics

There are filtering and sorting methods available for Stats objects in order to make the return data more useful.

It's important to profile before beginning to optimise as the source of slow runtime is not always obvious, or it can be misleading. By profiling function calls you can see exactly which functions are responsible for slow runtime.