How can I create a New numpy array of matrix dimensions 1x3,1x3 and 1x1 using an already existing 2D array containing 7 elements?

import numpy as np

def calculate(lst):
    arr = np.array(lst)
    print(arr.shape)
    newarr = arr.reshape((3, 3))
    mean = np.mean(newarr)
    rowmeans = np.mean(newarr, axis=1)
    columnmeans = np.mean(newarr, axis=0)
    combined_array1 = rowmeans.reshape(1, 3)
    combined_array2 = columnmeans.reshape(1, 3)
    combined_array3 = np.array([[mean]])  # Reshape mean to (1, 1)
    combinedarray = np.concatenate((combined_array1, combined_array2, combined_array3), axis=0)
    print(combinedarray)

calculate([1, 2, 3, 4, 5, 6, 7, 8, 9])

I want my answer to be like this but whenever I try to concatenate the third arrray I shows an error
[[2. 5. 8.] [4. 5. 6.] [5.]]

I works fine when I use hstack but I want a set of three different matrices?
It says all the input array dimensions except for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 3 and the array at index 2 has size 1

I works fine when I use hstack but I want a set of three different matrices?
It says all the input array dimensions except for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 3 and the array at index 2 has size 1

This [[2. 5. 8.] [4. 5. 6.] [5.]] is a list structure not an array. Normal Numpy arrays are regular arrays so that they can be stored in memory with space efficiency, efficient indexing of the memory locations, and high speed operation.

You can put the three numpy arrays into a Python list if you want. To be clear, the list would contain just the references to the original arrays, and accessing that list would just be a way to indirectly access the original arrays.

If you really must store this list structure in a numpy array, then numpy "object" arrays do exist for storing reference links, but I think this upends the whole point of numpy arrays, and will probably cause problems for you later.

combinedlist = [combined_array1, combined_array2, combined_array3]
combinedarray = np.empty([3], dtype=object)
combinedarray[:] = combinedlist

The output will be this:

[array([[2., 5., 8.]]) array([[4., 5., 6.]]) array([[5.]])]

Returning a list:

In [56]: def calculate(lst):
...:     arr = np.array(lst)
...:     print(arr.shape)
...:     newarr = arr.reshape((3, 3))
...:     mean = np.mean(newarr)
...:     rowmeans = np.mean(newarr, axis=1)
...:     columnmeans = np.mean(newarr, axis=0)
...:     combined_array1 = rowmeans.reshape(1, 3)
...:     combined_array2 = columnmeans.reshape(1, 3)
...:     combined_array3 = np.array([[mean]])  # Reshape mean to (1, 1)
...:     combined = [combined_array1, combined_array2, combined_array3]
...:     return combined
...: 
...: alist = calculate([1, 2, 3, 4, 5, 6, 7, 8, 9])
(9,)

That is 3 array, with shape (1,3),(1,3),(1,1). Pay attention to the double brackets.

In [57]: alist
Out[57]: [array([[2., 5., 8.]]), array([[4., 5., 6.]]), array([[5.]])]

hstack joins them on the last axis, axis 1:

In [58]: np.hstack(alist)
Out[58]: array([[2., 5., 8., 4., 5., 6., 5.]])
In [59]: np.concatenate(alist, axis=1)
Out[59]: array([[2., 5., 8., 4., 5., 6., 5.]])

axis=0 doesn't work because axis 1 varies, 3,3,1:

In [60]: np.concatenate(alist, axis=0)
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[60], line 1
----> 1 np.concatenate(alist, axis=0)
ValueError: all the input array dimensions except for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 3 and the array at index 2 has size 1
In [63]: [i.shape for i in alist]
Out[63]: [(1, 3), (1, 3), (1, 1)]