python - numpy array concatenate: "ValueError: all the input arrays must have same number of dimensions"

26
votes

How to concatenate these numpy arrays?

first np.array with a shape (5,4)

[[  6487    400 489580      0]
 [  6488    401 492994      0]
 [  6491    408 489247      0]
 [  6491    408 489247      0]
 [  6492    402 499013      0]]

second np.array with a shape (5,)

[  16.   15.   12.  12.  17. ]

final result should be

[[  6487    400    489580    0   16]
 [  6488    401    492994    0   15]
 [  6491    408    489247    0   12]
 [  6491    408    489247    0   12]
 [  6492    402    499013    0   17]]

I tried np.concatenate([array1, array2]) but i get this error

ValueError: all the input arrays must have same number of dimensions

What am I doing wrong?

3
pythonnumpy
How the heck is that second array supposed to have shape (1,)? Is there some sort of weird thing with object arrays going on here? - user2357112 supports Monica
That is what i got when i run array2.shape - RaduS
Then your array is seriously messed up in some way, and you need to figure out what's going on. - user2357112 supports Monica
just before that i run this array2 = np.array(np.round(data[:,0]/20)) - RaduS

3 Answers

35
votes

To use np.concatenate, we need to extend the second array to 2D and then concatenate along axis=1 -

np.concatenate((a,b[:,None]),axis=1)

Alternatively, we can use np.column_stack that takes care of it -

np.column_stack((a,b))

Sample run -

In [84]: a
Out[84]: 
array([[54, 30, 55, 12],
       [64, 94, 50, 72],
       [67, 31, 56, 43],
       [26, 58, 35, 14],
       [97, 76, 84, 52]])

In [85]: b
Out[85]: array([56, 70, 43, 19, 16])

In [86]: np.concatenate((a,b[:,None]),axis=1)
Out[86]: 
array([[54, 30, 55, 12, 56],
       [64, 94, 50, 72, 70],
       [67, 31, 56, 43, 43],
       [26, 58, 35, 14, 19],
       [97, 76, 84, 52, 16]])

If b is such that its a 1D array of dtype=object with a shape of (1,), most probably all of the data is contained in the only element in it, we need to flatten it out before concatenating. For that purpose, we can use np.concatenate on it too. Here's a sample run to make the point clear -

In [118]: a
Out[118]: 
array([[54, 30, 55, 12],
       [64, 94, 50, 72],
       [67, 31, 56, 43],
       [26, 58, 35, 14],
       [97, 76, 84, 52]])

In [119]: b
Out[119]: array([array([30, 41, 76, 13, 69])], dtype=object)

In [120]: b.shape
Out[120]: (1,)

In [121]: np.concatenate((a,np.concatenate(b)[:,None]),axis=1)
Out[121]: 
array([[54, 30, 55, 12, 30],
       [64, 94, 50, 72, 41],
       [67, 31, 56, 43, 76],
       [26, 58, 35, 14, 13],
       [97, 76, 84, 52, 69]])
4
votes

There's also np.c_

>>> a = np.arange(20).reshape(5, 4)
>>> b = np.arange(-1, -6, -1)
>>> a
array([[ 0,  1,  2,  3],
       [ 4,  5,  6,  7],
       [ 8,  9, 10, 11],
       [12, 13, 14, 15],
       [16, 17, 18, 19]])                                                                                                                                   
>>> b                                                                                                                                                       
array([-1, -2, -3, -4, -5])                                                                                                                                 
>>> np.c_[a, b]
array([[ 0,  1,  2,  3, -1],          
       [ 4,  5,  6,  7, -2],                       
       [ 8,  9, 10, 11, -3],                      
       [12, 13, 14, 15, -4],                                
       [16, 17, 18, 19, -5]])
2
votes

You can do something like this.

import numpy as np

x = np.random.randint(100, size=(5, 4))
y = [16, 15, 12, 12, 17]

print(x)

val = np.concatenate((x,np.reshape(y,(x.shape[0],1))),axis=1)
print(val)

This outputs:

[[32 37 35 53]
 [64 23 95 76]
 [17 76 11 30]
 [35 42  6 80]
 [61 88  7 56]]

[[32 37 35 53 16]
 [64 23 95 76 15]
 [17 76 11 30 12]
 [35 42  6 80 12]
 [61 88  7 56 17]]