How can I import an array to python (numpy.arry) from a file and that way the file must be written if it doesn't already exist.
For example, save out a matrix to a file then load it back.
Add a comment
|
5 Answers
Checkout the entry on the numpy example list. Here is the entry on .loadtxt()
>>> from numpy import *
>>>
>>> data = loadtxt("myfile.txt") # myfile.txt contains 4 columns of numbers
>>> t,z = data[:,0], data[:,3] # data is 2D numpy array
>>>
>>> t,x,y,z = loadtxt("myfile.txt", unpack=True) # to unpack all columns
>>> t,z = loadtxt("myfile.txt", usecols = (0,3), unpack=True) # to select just a few columns
>>> data = loadtxt("myfile.txt", skiprows = 7) # to skip 7 rows from top of file
>>> data = loadtxt("myfile.txt", comments = '!') # use '!' as comment char instead of '#'
>>> data = loadtxt("myfile.txt", delimiter=';') # use ';' as column separator instead of whitespace
>>> data = loadtxt("myfile.txt", dtype = int) # file contains integers instead of floats
1 Comment
Ricardo Rod
hello, thanks for answering, I can only doubt as he defines the path of where the file
Another option is numpy.genfromtxt, e.g:
import numpy as np
data = np.genfromtxt("myfile.dat",delimiter=",")
This will make data a numpy array with as many rows and columns as are in your file
1 Comment
Monty _s Flying Circus
Why is this answer less upvoted? This solved my problem. It seems more flexible, as well
(I know the question is old, but I think this might be good as a reference for people with similar questions)
If you want to load data from an ASCII/text file (which has the benefit or being more or less human-readable and easy to parse in other software), numpy.loadtxt is probably what you want:
If you just want to quickly save and load numpy arrays/matrices to and from a file, take a look at numpy.save and numpy.load:
Comments
In Python, Storing a bare python list as a numpy.array and then saving it out to file, then loading it back, and converting it back to a list takes some conversion tricks. The confusion is because python lists are not at all the same thing as numpy.arrays:
import numpy as np
foods = ['grape', 'cherry', 'mango']
filename = "./outfile.dat.npy"
np.save(filename, np.array(foods))
z = np.load(filename).tolist()
print("z is: " + str(z))
This prints:
z is: ['grape', 'cherry', 'mango']
Which is stored on disk as the filename: outfile.dat.npy
The important methods here are the tolist() and np.array(...) conversion functions.
Comments
Have a look at SciPy cookbook. It should give you an idea of some basic methods to import /export data.
If you save/load the files from your own Python programs, you may also want to consider the Pickle module, or cPickle.
1 Comment
dwf
Pickling is inappropriate for arrays - while you can do it, it will be slow as hell. Use np.save() to save in the .npy format or np.savez() to save a zipped archive of several arrays.