Reshaping and stacking

2 min read 7 runnable examples Quiz included

reshape

Python

import numpy as np

a = np.arange(1, 13)            # 12 numbers in a row
print(a.reshape(3, 4))          # 3 rows of 4
print(a.reshape(2, 6))
print(a.reshape(2, 2, 3))       # 2 blocks of 2 rows by 3 columns

reshape arranges the same elements into a new shape, filling row by row. The total must stay the same: 12 elements can be 3 × 4 or 2 × 6, but not 5 × 3, which raises an error.

Let NumPy work out one dimension: -1

Python

import numpy as np

a = np.arange(12)
print(a.reshape(4, -1).shape)    # (4, 3): -1 means "whatever fits"
print(a.reshape(-1, 6).shape)    # (2, 6)
print(a.reshape(-1).shape)       # (12,): back to one dimension

Flattening

Python

import numpy as np

grid = np.array([[1, 2, 3],
                 [4, 5, 6]])
print(grid.flatten())    # a 1D copy
print(grid.ravel())      # a 1D view where possible (no copy)

Transposing

Python

import numpy as np

m = np.array([[1, 2, 3],
              [4, 5, 6]])
print(m.T)              # rows become columns
print(m.shape, "->", m.T.shape)

.T swaps the axes: a 2 × 3 array becomes 3 × 2. It returns a view, so it costs nothing to compute.

Adding an axis

Python

import numpy as np

v = np.array([1, 2, 3])
print(v.shape)                      # (3,)
print(v[np.newaxis, :].shape)       # (1, 3): a row
print(v[:, np.newaxis].shape)       # (3, 1): a column
print(v.reshape(-1, 1))             # the same column, using reshape

A 1D array is neither a row nor a column. Adding an axis makes it one or the other, which matters for broadcasting, the next lesson.

Joining arrays

Python

import numpy as np

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

print(np.concatenate([a, b]))           # stacked vertically (axis 0)
print(np.concatenate([a, b], axis=1))   # side by side (axis 1)
print(np.vstack([a, b]).shape)          # (4, 2): the same as axis 0
print(np.hstack([a, b]).shape)          # (2, 4): the same as axis 1

The arrays must match along every axis except the one you join on.

Splitting arrays

Python

import numpy as np

a = np.arange(12)
first, second, third = np.split(a, 3)          # three equal parts
print(first, second, third)

m = np.arange(16).reshape(4, 4)
top, bottom = np.split(m, 2)                   # split the rows in half
left, right = np.split(m, 2, axis=1)           # split the columns in half
print(top)
print(right)

Finished this lesson?Mark it complete to earn +20 XP and move your progress on.