Extend numpy array

Question

I have two 2d numpy arrays like

array2D_1 = np.arange(9).reshape(3,3)
array2D_2 = np.arange(10,19).reshape(3,3)

Is it possible to append the second array to the first as just one column?

To concat two numpy arrays I have tryed the following function:

def _concat_two_features(self, array2D_1, array2D_2):
    return np.concatenate((array2D_1, array2D_2), axis=1)

and also

def _append_two_features(self, array2D_1, array2D_2):
        return np.append((array2D_1, array2D_2), axis=1)

Both result in the same array

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

What I would like to have is:

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

I also tried something like this:

np.concatenate((array2D_1,[array2D_2]),axis=1)

But this doesn't the operation eather.

ValueError: all the input arrays must have same number of dimensions, but the array at index 0 has 2 dimension(s) and the array at index 1 has 3 dimension(s)

Thanks in advance

With axis, np.append just does concatenate. Look at its code — hpaulj
– hpaulj, Commented Sep 8, 2021 at 14:43
@some_name.py I have a multi dimensional feature which need to be considered as single value — Jürgen K.
– Jürgen K., Commented Sep 8, 2021 at 14:49
@hpaulj what do you want to say? the output shows your answer already — Jürgen K.
– Jürgen K., Commented Sep 8, 2021 at 14:49
What you seek is not a multidimensional numeric array. concatenate can't do it. — hpaulj
– hpaulj, Commented Sep 8, 2021 at 14:53

some_name.py · Accepted Answer · 2021-09-08 14:50:57Z

2

Does this help?

import numpy as np

array2D_1 = np.arange(9).reshape(3,3)
array2D_2 = np.arange(10,19).reshape(3,3)

out = np.empty(shape=(3,4),dtype=np.object)
out[:,:3] = array2D_1

out[:,3] = list(array2D_2)

Output:

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

answered Sep 8, 2021 at 14:50

some_name.py

8377 silver badges17 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Jürgen K. Over a year ago

This is quiet perfect! Thx

hpaulj · Accepted Answer · 2021-09-08 15:09:06Z

1

OK, there is a way to use concatenate, but you have to work with object dtype arrays. This can't be done with numeric arrays.

In [6]: x=np.empty(array2D_2.shape[0],object); x[:]=list(array2D_2);
In [7]: x
Out[7]: 
array([array([10, 11, 12]), array([13, 14, 15]), array([16, 17, 18])],
      dtype=object)
In [8]: np.concatenate((array2D_1.astype(object), x[:,None]), axis=1)
Out[8]: 
array([[0, 1, 2, array([10, 11, 12])],
       [3, 4, 5, array([13, 14, 15])],
       [6, 7, 8, array([16, 17, 18])]], dtype=object)

answered Sep 8, 2021 at 15:09

hpaulj

233k14 gold badges260 silver badges392 bronze badges

1 Comment

Jürgen K. Over a year ago

Thx, great idea. However, I will accept the solution that was proposed preveously, since it is more readable

Collectives™ on Stack Overflow

Extend numpy array

2 Answers 2

1 Comment

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related