A (very basic) look at the (statistically significant) most performant python operation options

Obviously, the outcomes are very dependent on the test cases...

Notation

(case1), case4 >> case2, case3

i.e., (case 1 - with min point estimate exectution mean), was tested against 2,3,4 and found to be significantly faster (>>) than 2 and 3, but not 4

Reduction operations:

takeaway: **use for loops for reducing**

• Test: Return the maximum value in a randomly shuffled list of 1e6 numbers (for loop) >> reduce()

Summing a list:

takeaway: **use sum()**

• Test: Return the sum of a randomly shuffled list of 1e6 numbers (sum()) >> reduce()

Casting iterators/ generators to lists

takeaway: **use list comprehensions in preference to casting iterators**

• Test: iterate through range(1000000), introducing each element to the iterable. non-list iterables are then cast to list. (list comp), list(generator) >> for loop, while, list(map)

• Test: Filtering (list comp) >> list(filter), list(generator)

Creating iteratables and using in subsequent list comprehensions

takeaway: **use intermediate generator objects**

• Test: assemble an intermediate iterator of 1e6 elements, and use as a basis for a list comprehension. (generator) >> list comp., map