Transforming a list to a column vector in Python

Question

I was wondering if there's a function that takes a list X as input and outputs a column vector corresponding to X?

(I looked around and I suspect it might be: np.matrix(X).T )

Bonanza

Are you using numpy? – Nubbin

Yes I am using numpy – Bonanza

np.reshape(X, (len(X), 1)) - It'll also accept X even if it's a list or tuple – Killie

@drooze. -1 probably easier than len(X) here. – Nubbin

@Mkach. I've edited your tags to reflect that. This is a very common question so I'm glad you asked it. It's probably a duplicate but I'm too lazy to look so just posted an answer instead. – Nubbin

The basic point is that np.array([1,2,3,4]), makes a 1d array, shape (4,). Often that's enough. A column vector has shape (4,1). There are lots of way of adding that dimension to the 1d array. – Ernesternesta

Answer 1

I don't think there is a function, but there is a dedicated object, np.c_

>>> X = list('hello world')
>>> X
['h', 'e', 'l', 'l', 'o', ' ', 'w', 'o', 'r', 'l', 'd']
np.c_[X]
array([['h'],
       ['e'],
       ['l'],
       ['l'],
       ['o'],
       [' '],
       ['w'],
       ['o'],
       ['r'],
       ['l'],
       ['d']], dtype='<U1')

You can also do (note the extra square brackets)

>>> np.array([X]).T

Gauge

Answer 2

If you want to convert a Python list into a numpy column vector, you can use the ndmin argument to the array conductor:

col_vec = np.array(X, ndmin=2)

Simply constructing an array will give you a 1D array, which can not be directly transposed:

a = np.array(X)
a is a.T

There are a couple of ways to transform a 1D vector into a column though:

col_vec = a.reshape(-1, 1)

and

col_vec = a[np.newaxis, :]

are fairly common idioms after your list is already an array.

col_vec = np.reshape(X, (-1, 1))

and

col_vec = np.expand_dims(X, -1)

will work even on a raw list.

P.S. Stay away from np.matrix it's very outdated and limited, if not outright deprecated.

Nubbin

Answer 3

A list is a native type of Python, whereas a numpy array is a numpy object. You will need to convert your list to a numpy array first. You can do something like the following.

x = list(range(5))
print(x)

[0, 1, 2, 3, 4]

x_vector = np.asarray(x)

array([0, 1, 2, 3, 4])

Now, Python does not know whats the difference between a row vector and a column vector, that is up to how you will use the vector. The current vector is 1x5. If you want 5x1, you can take the transpose by either

x_vector.T

or

np.transpose(x_vector)

However, since this is a 1D matrix, the transpose is the same as the non-transposed vector in memory.

Estimative

As you guessed, transpose won't work correctly. You can't swap two dims when there aren't two dims to swap. – Nubbin

Answer 4

similar but different:

mylist=[2,3,4,5,6]

method1:

np.array([[i] for i in mylist])

output:

array([[2],
       [3],
       [4],
       [5],
       [6]])

method 2:

np.array(mylist).reshape(len(mylist),1)

output:
    array([[2],
           [3],
           [4],
           [5],
           [6]])

Ladyinwaiting