Friday, August 16, 2013

Numpy vs. cPickles (Python, ofc)

I've been using cPickels for storing data into a Python-friendly format for some time. See my earlier blog post for more on cPickles.
http://combichem.blogspot.dk/2013/02/saving-into-data-into-cpickle-format-in.html

I have also been using Numpy's save function to do the same thing. numpy.save() and numpy.load() is so much simpler, however. I really recommend that people use numpy.save() and numpy.load() over cPickles for most purposes. It is so much more simple.

I always thought a cPickle was much, much faster than Numpy, but I guess I was wrong, according to this stackoverflow I just saw. Below are loading and saving times for a large array. Practically no difference between Numpy and cPickles!

Source: http://stackoverflow.com/questions/16833124/pickle-faster-than-cpickle-with-numeric-data








To save an array, a list or dictionary or whatever called my_array into my_file.npy:

  numpy.save("my_file", my_array)

Note that Numpy appends .npy to the filename automatically.


To load the stored data simply:

  my_array = numpy.load("my_file.npy")

 Really py-fragging-thonicly easy!

2 comments:

  1. It has been a real frightening difficulty in my position, but witnessing a expert style you dealt with that took me to weep for happiness. Check shabbat candles for best candles.

    ReplyDelete
  2. Good information provided by thanks keep updating
    python online training

    ReplyDelete