How do I use numpy's stack, vstack, and hstack?

This post contains examples of formatting numpy arrays with stack, vstack, and hstack.

import numpy as np

Sample data

Let’s start with some normal Python lists to start.

arrays = [[i, i+1] for i in range(0, 6, 2)]
arrays

[[0, 1], [2, 3], [4, 5]]

numpy.stack

https://docs.scipy.org/doc/numpy/reference/generated/numpy.stack.html

numpy.stack is the most general of the three methods, offering an axis parameter for specifying which way to put the arrays together.

With axis 0, we end up with a shape similar to what our original Python lists were in.

stack_axis_zero = np.stack(arrays, axis=0)
stack_axis_zero, stack_axis_zero.shape

(array([[0, 1],
        [2, 3],
        [4, 5]]), (3, 2))

Notice how this is similar to the standard numpy.array creation from Python lists function.

np.array(arrays), np.array(arrays).shape

(array([[0, 1],
        [2, 3],
        [4, 5]]), (3, 2))

numpy.stack is more abstract, however, and offers an axis change as well.

stack_axis_one = np.stack(arrays, axis=1)
stack_axis_one, stack_axis_zero.shape

(array([[0, 2, 4],
        [1, 3, 5]]), (3, 2))

numpy.vstack

numpy.vstack implies a vertical stack, and we end up with a similar result as above.

vertical_stack = np.vstack(arrays)
vertical_stack, vertical_stack.shape

(array([[0, 1],
        [2, 3],
        [4, 5]]), (3, 2))

Notice how this is similar to the more abstract numpy.stack method, operating on axis 0.

s = np.stack(arrays, axis=0)
s, s.shape

(array([[0, 1],
        [2, 3],
        [4, 5]]), (3, 2))

Which is also similar to creating the expected shape from numpy.array, when operating on lists.

a = np.array(arrays)
a, a.shape

(array([[0, 1],
        [2, 3],
        [4, 5]]), (3, 2))

numpy.hstack

nummpy.hstack is a little different, reducing the dimension and appending the arrays into one.

horizontal_stack = np.hstack(arrays)
horizontal_stack, horizontal_stack.shape

(array([0, 1, 2, 3, 4, 5]), (6,))

But it can also work on higher order dimensions.

a = np.array([[1],[2],[3]])
b = np.array([[4],[5],[6]])
c = np.array([[7],[8],[9]])

np.hstack([a, b, c])

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