CODEWITHSUNDEEP
pythonjupyter-notebookexecution-timetimeit
Raval Sahil
Raval Sahil

Reputation: 1

I'm using %%timeit in jupyter notebooks to get the execution time of my cell

whenever I run the cell the time is coming in microseconds, I want the time in seconds how can I achieve this?

enter image description here

I want time in seconds or how can I get the result in another time parameter like second or minute

Upvotes: -1

Views: 796

Answers (1)

furas
furas

Reputation: 142631

Doc for %timeit

If you put code in def function() then you can use %time (with single %)
with option -o function() to get result in variable

variable = %time -o function()

It doesn't work with double %%. It raises error when it sees %% after =.

For example (all in one cell)

def function():
    for x in range(1_000_000):
        x**2

variable = %timeit -o function()

print(f'{variable.average:.3} s')

But it still displays original result before your print()

EDIT:

for %% you can get result in _ in next cell

%%timeit -o
for x in range(1_000_000):
    x**2

next cell

print(f'{_.average:.3} s')

But it still displays original result before first cell

You can use %%capture to hide original result

%%capture
%%timeit -o
for x in range(1_000_000):
    x**2

Based on documentation for %%capture you can even use it to assign output to variable but it will be string, not object with .average, so it will be hard to format it. And it doesn't work for me :/

%%capture original_output
%%timeit -o
for x in range(1_000_000):
    x**2

next cell

print(f'{_.average:.3} s')
print(f'original output: {original_output}')

If you use print( dir(variable) ) then you see what other values you can get from variable and you can use them to format output on your own.

To skip names which start with _ (like __class__) you can use list comprehension:

[var for var in dir(variable) if not var.startswith('_')]

['all_runs',
 'average',
 'best',
 'compile_time',
 'loops',
 'repeat',
 'stdev',
 'timings',
 'worst']

EDIT:

Here notebook on nbviewer.org created by @Wayne (see comment below).

To make sure I add also screenshot:

enter image description here

Upvotes: 2

Related Questions