[FIXED] Convert numpy array in column vector

Issue

I am new to python programming so excuse me if the question may seem silly or trivial.

So for a certain function I need to do a check in case x is a vector and (in that case) it must be a column vector. However I wanted to make it so that the user could also pass it a row vector and turn it into a column vector.

However at this point how do I make it not do the check if x is a scalar quantity.

I attach a code sketch to illustrate the problem. Thanks in advance to whoever responds

import numpy as np

x = np.arange(80, 130, 10)
# x: float = 80.0
print('array before:\n', x)

if x is not np.array:
    x = np.array(x).reshape([len(x), 1])

print('array after:\n', x)

Solution

You can check the type using if not isinstance(x, np.ndarray). After that check you can check the number of dimension of the array and convert to a columnar array.

In the code below, first if block checks and converts to an array. Then we get how many dimensions are missing. I.e. a scalar value is missing 2 dimensions, a flat list is missing 1 dimension. Finally we iterate over the number of missing dimensions up-converting the array's dimensions with each step.

def to_column_array(x):
    if not isinstance(x, np.ndarray):
        x = np.array(x)

    missing_dims = 2 - x.ndim
    if missing_dims < 0:
        raise ValueError('You array has too many dimensions')

    for _ in range(missing_dims):
        x = x.reshape(-1, 1)
    return x

So if a user has just a number, this convert it to an array of size (1, 1).

to_column_array(10)
# returns:
array([[10]])

Passing in a list or list of lists converts to an array as well.

to_column_array([3, 6, 9])
# returns:
array([[3],
       [6],
       [9]])

to_column_array([[1, 2], [3, 4], [5, 6])
# returns:
array([[1, 2],
       [3, 4],
       [5, 6]])

Answered By - James