Loading a file into a numpy array with python

Question

So I'm very green with Python and am trying to learn by replicating some matlab code I've written. I have a part where, in matlab, I load a data file that's tab-delimited. The syntax

x = load(data.txt)

Takes the tab delimited data and put them into cells of a matrix labeled x.

Is there a way to do this in python, but with comma-delimited data?

Answer 1

There are several methods, choose one that is most suitable for your application.

If you are working with numpy, it may be a good idea to use the numpy's load, loadtxt, fromfile or genfromtxt functions, because your file will be loaded into a suitable structure, after the preprocessing.

But if you are not about to work with numpy (or any other big library which has some file loading functionalities), it would be an overkill using it just for loading a file ... Consider using built-in python functions, or the csv module from the standard library instead ... It will be much more flexible, and way smoother.

Here is how, with examples using file.txt (values of each rows are separated with tabs):

1   2   3   4
7   8   9   10  11  12
13  14  15

python built-in

No module to import, pretty easy, flexible, a good option for most situations, imho.

Loading the file in binary mode for reading (rb flags) in a table (list of lists of values, separated in the file with tabs) with only built-in functions:

>>> file = open('file.txt', 'rb')
>>> table = [row.strip().split('\t') for row in file]

csv

The csv module from the standard library is pretty straightforward as well.

Note that altough CSV means Comma Separated Values, there is actually no standard and you can choose any delimiter you want. Therefore CSV stands for all cells-oriented or table-like files.

Loading the file in binary mode for reading (rb flags) in a table (list of lists of values, separated in the file with tabs) with the csv reader:

>>> import csv
>>> file = open('file.txt', 'rb')
>>> data = csv.reader(file, delimiter='\t')
>>> table = [row for row in data]

Accessing the cells

The table has been loaded similarly with the two previous examples, and the data of the table can be accessed like table[row][col]:

>>> table
[['1', '2', '3', '4'], ['7', '8', '9', '10', '11', '12'], ['13', '14', '15']]    
>>> table[0]
['1', '2', '3', '4']
>>> table[1][2]
9

Answer 2

The numpy.loadtxt function reads data from an ASCII file into a numpy array. The string used to separate values can be defined with the delimiter argument:

numpy.loadtxt('data.txt', delimiter=',')

For more complicated cases, the numpy.genfromtxt function is a very good alternative.

Answer 3

There is a csv module in the standard library.

See the documentation here

>>> import csv
>>> spamReader = csv.reader(open('eggs.csv', 'rb'), delimiter=' ', quotechar='|')
>>> for row in spamReader:
...     print ', '.join(row)
Spam, Spam, Spam, Spam, Spam, Baked Beans
Spam, Lovely Spam, Wonderful Spam

Answer 4

See the csv module (specifically the reader class) and/or numpy.loadtxt function.

Answer 5

If you're using Python for MATLAB-like purposes you're going to want to be using NumPy (and scipy); in particular, you should read NumPy for MATLAB Users.

If you have comma-delimited data, you can use numpy.loadtxt to read it (after installing numpy, of course):

$ cat matrix.csv 
1,2,3
4,5,6
7,8,9

and then

>>> import numpy as np
>>> m = np.loadtxt("matrix.csv", delimiter=",")
>>> m
array([[ 1.,  2.,  3.],
       [ 4.,  5.,  6.],
       [ 7.,  8.,  9.]])
>>> np.matrix(m)
matrix([[ 1.,  2.,  3.],
        [ 4.,  5.,  6.],
        [ 7.,  8.,  9.]])