If I have a NumPy array, for example 5x3, is there a way to unpack it column by column all at once to pass to a function rather than like this: my_func(arr[:, 0], arr[:, 1], arr[:, 2])?
Kind of like *args for list unpacking but by column.
3 Answers
You can unpack the transpose of the array in order to use the columns for your function arguments:
my_func(*arr.T)
Here's a simple example:
>>> x = np.arange(15).reshape(5, 3)
array([[ 0, 5, 10],
[ 1, 6, 11],
[ 2, 7, 12],
[ 3, 8, 13],
[ 4, 9, 14]])
Let's write a function to add the columns together (normally done with x.sum(axis=1) in NumPy):
def add_cols(a, b, c):
return a+b+c
Then we have:
>>> add_cols(*x.T)
array([15, 18, 21, 24, 27])
NumPy arrays will be unpacked along the first dimension, hence the need to transpose the array.
Comments
numpy.split splits an array into multiple sub-arrays. In your case, indices_or_sections is 3 since you have 3 columns, and axis = 1 since we're splitting by column.
my_func(numpy.split(array, 3, 1))
Comments
I guess numpy.split will not suffice in the future. Instead, it should be
my_func(tuple(numpy.split(array, 3, 1)))
Currently, python prints the following warning:
FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use
arr[tuple(seq)]instead ofarr[seq]. In the future this will be interpreted as an array index,arr[np.array(seq)], which will result either in an error or a different result.
1 Comment
Michael B. Currie
This gave me the answer I needed: wrap the numpy array in a
tuple().
np.hsplit(x,x.shape[1])