Differences between Flatten() and Ravel() Numpy Functions

27 July 2024

1

We have two similar kinds of ways to convert a ndarray to a 1D array of Flatten() and Ravel() Numpy function in Python programming language.

Example of Flatten() Numpy function

Here, we will create a Numpy array, and then by using flatten function we have changed the element in the flattened 1D NumPy array.

Python3

import numpy as np
 
# Create a numpy array
a = np.array([(1,2,3,4),(3,1,4,2)])
 
# Let's print the array a
print ("Original array:\n ", a)
 
# Convert nd array to 1D array
c = a.flatten()
 
# Flatten passes copy of
# original array to 'c'
print ("\nFlatten array using flatten: ", c)

Output:

Original array:
  [[1 2 3 4]
 [3 1 4 2]]

Flatten array using flatten:  [1 2 3 4 3 1 4 2]

The original 2D array is unchanged even if the value of the second member in the flattened 1D Numpy array was changed. This demonstrates that flatten() does return a copy of the Numpy input array.

Python3

c[2] = 33
 
print ("Flatten array: ", c)
 
# Let's print the array a
print ("\nOriginal array:\n ", a)

Output:

Flatten array:  [ 1  2 33  4  3  1  4  2]

Original array:
  [[1 2 3 4]
 [3 1 4 2]]

Example of ravel() Numpy function

Here, we will create a Numpy array, and then by using the ravel() function we have changed the element in the flattened 1D NumPy array.

Python3

import numpy as np
 
# Create a numpy array
a = np.array([(1,2,3,4),(3,1,4,2)])
 
# Let's print the array a
print ("Original array:\n ", a)
 
ra = np.ravel(a)
print ("\nFlatten array using ravel: ", ra)

Output:

`Original array:
  [[1 2 3 4]
 [3 1 4 2]]

Flatten array using ravel:  [1 2 3 4 3 1 4 2]

As we can see, The original 2D array is changed if the value of the second member in the flattened 1D Numpy array was changed. This demonstrates that ravel() returns a shallow copy of the Numpy input array.

Python3

ra[3] = 44
 
print ("Flatten array: ", ra)
 
# Let's print the array a
print ("\nOriginal array:\n ", a)

Output:

Flatten array:  [ 1  2  3 44  3  1  4  2]

Original array:
  [[ 1  2  3 44]
 [ 3  1  4  2]]