NumPy: Error While Concatenation - zero-dimensional arrays cannot be concatenated

Question

I am trying to concat two valid array via np.concat() method.

My code:

print X_train.shape, train_names.shape
X_train = np.concatenate([train_names,X_train], axis=0)

The output:



    (3545, 93355) (3545, 692)


    ValueError                                Traceback (most recent call last)
    <ipython-input-58-59dc66874663> in <module>()
      1 print X_train.shape, train_names.shape
    ----> 2 X_train = np.concatenate([train_names,X_train], axis=0)
      

    ValueError: zero-dimensional arrays cannot be concatenated

As you can see, the shapes of arrays align, still I am getting this weird error. Why?

EDIT: I have tried with axis=1 as well. Same result

EDIT 2: Eqauted data types using .astype(np.float64). Same result.

I would also check the dtype of the inputs. Something in the input to the np.concatenate is zero-dimensional. We can't see it, but you will if you add enough diagnostic prints. — hpaulj
– hpaulj, Commented Apr 5, 2016 at 2:49
I tried equating dtypes. Still same result. Maybe it is a bug with my version of Numpy 1.10.4. Plausible? — silent_grave
– silent_grave, Commented Apr 5, 2016 at 3:18
By sparse do you mean lots of zeros, or the scipy sparse matrix type? Since it works for smaller arrays, I suspect a memory or indexing issue. We need to look for other concatenate error questions. — hpaulj
– hpaulj, Commented Apr 5, 2016 at 3:56
If your arrays are scipy sparse don't use np.concatenate. Only use sparse functions. — hpaulj
– hpaulj, Commented Apr 5, 2016 at 4:12

hpaulj · Accepted Answer · 2016-04-05 04:14:17Z

Applying np.concatenate to scipy sparse matrices produces this error:

In [162]: from scipy import sparse
In [163]: x=sparse.eye(3)
In [164]: x
Out[164]: 
<3x3 sparse matrix of type '<class 'numpy.float64'>'
    with 3 stored elements (1 diagonals) in DIAgonal format>
In [165]: np.concatenate((x,x))
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-165-0b67d0029ca6> in <module>()
----> 1 np.concatenate((x,x))

ValueError: zero-dimensional arrays cannot be concatenated

There are sparse functions to do this:

In [168]: sparse.hstack((x,x)).A
Out[168]: 
array([[ 1.,  0.,  0.,  1.,  0.,  0.],
       [ 0.,  1.,  0.,  0.,  1.,  0.],
       [ 0.,  0.,  1.,  0.,  0.,  1.]])
In [169]: sparse.vstack((x,x)).A
Out[169]: 
array([[ 1.,  0.,  0.],
       [ 0.,  1.,  0.],
       [ 0.,  0.,  1.],
       [ 1.,  0.,  0.],
       [ 0.,  1.,  0.],
       [ 0.,  0.,  1.]])

xthestreams · Accepted Answer · 2016-04-05 02:06:05Z

1

Pass the arrays as a tuple rather than a list. X_train = np.concatenate((train_names,X_train), axis=0)

answered Apr 5, 2016 at 2:06

xthestreams

1695 bronze badges

1 Comment

hpaulj Over a year ago

In this context list and tuples are treated the same.

Ashish Gupta · Accepted Answer · 2019-01-02 09:50:35Z

1

In my case, the problem was due to sparse nature of matrix returned by LabelEncoder()

so this line gave me an error:

np.append(airlineTrain, train_transformed,axis =1)

To fix it, I used this:

np.append(airlineTrain.toarray(), train_transformed.toarray(),axis =1 )

Alternately, you may be using NLTK, where the sparse storage of matrices can be converted by using todense()

answered Jan 2, 2019 at 9:50

Ashish Gupta

1852 silver badges13 bronze badges

Collectives™ on Stack Overflow

NumPy: Error While Concatenation - zero-dimensional arrays cannot be concatenated

3 Answers 3

Comments

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related