How to randomly insert NaN in a matrix with numpy in python ?

Inserting NaN values into a matrix is a relatively straightforward task when using numpy in python. Examples:

Create a matrix with numpy

First, we must create the array that will hold our matrix (or alternatively load the matrix from a file):

import numpy as np

A = np.random.uniform(10,80, size=(4,6))

print(A)

Ouput

[[52.34830542 43.7300926  65.65912419 74.47707968 47.7363097  31.78605372]
 [49.41123686 19.82268971 59.91408598 40.86920833 53.23834812 37.93559161]
 [16.18419498 42.49772722 76.53306408 30.90572765 38.15287236 72.44956349]
 [46.15969878 52.39722864 71.97596547 70.40800518 44.63824773 35.43923044]]

Randomly insert NaN

Then, we can use np.random.choice() to randomly select elements in the array and replace them with NaN values. For example:

n = 6

index = np.random.choice(A.size, n, replace=False)

A.ravel()[index] = np.nan

print(A)

Ouput

[[52.34830542 43.7300926  65.65912419         nan 47.7363097          nan]
 [        nan 19.82268971         nan 40.86920833 53.23834812 37.93559161]
 [16.18419498 42.49772722         nan 30.90572765 38.15287236 72.44956349]
 [46.15969878 52.39722864 71.97596547         nan 44.63824773 35.43923044]]

Another example

Note:

print(type(np.nan))

gives

 <class 'float'>

An example

import numpy as np

A = np.random.randint(10,80, size=(5,2))

A = A * 1.0

print(A)

returns

[[52. 55.]
[14. 33.]
[19. 50.]
[67. 37.]
[16. 72.]]

and

n = 3

index = np.random.choice(A.size, n, replace=False)

A.ravel()[index] = np.nan

print(A)

returns for example

[[52. 55.]
 [14. 33.]
 [19. nan]
 [67. nan]
 [16. nan]]

References