Menu
I want to turn my array of array into just a single array. From something like :
array([ array([[0, 0, 0, ..., 1, 0, 0],
[0, 1, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 2, 0, 0],
...,
array([[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 8, 0, 2],
...,
[0, 0, 0, ..., 0, 0, 0],
[1, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 1, 0, 0]], dtype=uint8)], dtype=object)
which has size (10,) to just the 3D numpy array which is of size (10,518, 32)
array([[[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]]], dtype=uint8)
I've tried converting everything into a list then do np.asarray and also tried defining everything as the same dtype=uint8 but I couldn't get it into the 3D form.
900
Use numpy. concatenate() to merge the content of two or multiple arrays into a single array. This function takes several arguments along with the NumPy arrays to concatenate and returns a Numpy array ndarray. Note that this method also takes axis as another argument, when not specified it defaults to 0.
NumPy's concatenate function can be used to concatenate two arrays either row-wise or column-wise. Concatenate function can take two or more arrays of the same shape and by default it concatenates row-wise i.e. axis=0. The resulting array after row-wise concatenation is of the shape 6 x 3, i.e. 6 rows and 3 columns.
You can use numpy. append() function to add an element in a NumPy array. You can pass the NumPy array and multiple values as arguments to the append() function. It doesn't modify the existing array but returns a copy of the passed array with given values added.
flatten() function return a copy of the array collapsed into one dimension. Parameters : order : [{'C', 'F', 'A', 'K'}, optional] 'C' means to flatten in row-major (C-style) order. 'F' means to flatten in column-major (Fortran- style) order.
np.concatenate should do the trick:
Make an object array of arrays:
In [23]: arr=np.empty((4,),dtype=object)
In [24]: for i in range(4):arr[i]=np.ones((2,2),int)*i
In [25]: arr
Out[25]:
array([array([[0, 0],
[0, 0]]), array([[1, 1],
[1, 1]]),
array([[2, 2],
[2, 2]]), array([[3, 3],
[3, 3]])], dtype=object)
In [28]: np.concatenate(arr)
Out[28]:
array([[0, 0],
[0, 0],
[1, 1],
[1, 1],
[2, 2],
[2, 2],
[3, 3],
[3, 3]])
Or with a reshape:
In [26]: np.concatenate(arr).reshape(4,2,2)
Out[26]:
array([[[0, 0],
[0, 0]],
[[1, 1],
[1, 1]],
[[2, 2],
[2, 2]],
[[3, 3],
[3, 3]]])
In [27]: _.shape
Out[27]: (4, 2, 2)
concatenate effectively treats its input as a list of arrays. So it works regardless of whether this is an object array, a list, or 3d array.
This can't be done simply with a reshape. arr is an array of pointers - pointing to arrays located elsewhere in memory. To get a single 3d array, all of the pieces will have to be copied into one buffer. That's what concatenate does - it creates a large empty file, and copies each array, but it does it in compiled code.
np.array does not change it:
In [37]: np.array(arr).shape
Out[37]: (4,)
but treating arr as a list of arrays does work (but is slower than the concatenate version - array analyses its inputs more).
In [38]: np.array([x for x in arr]).shape
Out[38]: (4, 2, 2)
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With
