dataFrame python's really easy to use, but obviously only single-core computing
Use pandas, when you run the following row:
# Standard apply
df.apply(func)
Get this CPU usage:
Even if the computer has a plurality of CPU, only a fully dedicated to the calculation.
Recently recommended by the group of friends began to find the accelerator, really cow fork! ! ! You can truly experience the nuclear stand-alone with a python can also be fully open, off the eight-core thrill! !
[Pandaral·lel] The idea is to calculate the pandas distributed across all available CPU on a computer, to significantly improve the speed.
installation:
$ pip install pandarallel [--user]
Import and initialization:
Import:
from pandarallel import pandarallel
Initialization
pandarallel.initialize()
Usage is very simple:
usage:
Use with pandas DataFrame function func simple to apply and use embodiments of df, simply replace the parallel_apply classic apply.
# Standard pandas apply
df.apply(func)
# Parallel apply
df.parallel_apply(func)
