How to remove a column in a numpy array?

Question

Imagine we have a 5x4 matrix. We need to remove only the first dimension. How can we do it with numpy?

array([[  0.,   1.,   2.,   3.],
       [  4.,   5.,   6.,   7.],
       [  8.,   9.,  10.,  11.],
       [ 12.,  13.,  14.,  15.],
       [ 16.,  17.,  18.,  19.]], dtype=float32)

I tried:

arr = np.arange(20, dtype=np.float32)
matrix = arr.reshape(5, 4)
new_arr = numpy.delete(matrix, matrix[:,0])
trimmed_matrix = new_arr.reshape(5, 3)

It looks a bit clunky. Am I doing it correctly? If yes, is there a cleaner way to remove the dimension without reshaping?

Answer 1

If you want to remove a column from a 2D Numpy array you can specify the columns like this

to keep all rows and to get rid of column 0 (or start at column 1 through the end)

a[:,1:]

another way you can specify the columns you want to keep ( and change the order if you wish) This keeps all rows and only uses columns 0,2,3

a[:,[0,2,3]]

The documentation on this can be found here

And if you want something which specifically removes columns you can do something like this:

idxs = list.range(4)
idxs.pop(2) #this removes elements from the list
a[:, idxs]

and @hpaulj brought up numpy.delete()

This would be how to return a view of 'a' with 2 columns removed (0 and 2) along axis=1.

np.delete(a,[0,2],1)

This doesn't actually remove the items from 'a', it's return value is a new numpy array.