# How to randomly select elements of an array with numpy in python ?

Published: November 26, 2019

Examples of how to randomly select elements of an array with numpy in python:

### Randomly select elements of a 1D array using choice()

Lets create a simple 1D array with 10 elements:

>>> import numpy as np
>>> data = np.arange(10)
>>> data
array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])


A = \left( \begin{array}{ccc}
0 & 1& 2& 3& 4& 5& 6& 7& 8& 9
\end{array}\right)

To select randomly n elements, a solution is to use choice(). Example of how to select randomly 4 elements from the array data:

>>> np.random.choice(data,4)
array([9, 6, 2, 9])


returns for example

A = \left( \begin{array}{ccc}
9 & 6 & 2 & 9
\end{array}\right)

Another example, with n = 5

>>> for i in range(10):
...     np.random.choice(data,5)
... 
array([3, 4, 0, 8, 4])
array([3, 0, 0, 3, 6])
array([5, 1, 2, 0, 9])
array([5, 8, 6, 0, 1])
array([4, 0, 9, 4, 2])
array([9, 6, 3, 9, 9])
array([9, 5, 1, 2, 7])
array([9, 7, 6, 4, 5])
array([6, 8, 5, 5, 9])
array([8, 9, 5, 5, 6])


### Random sampling without replacement

To do random sampling without remplacement, just add the option "replace = False":

>>> for i in range(10):
...     np.random.choice(data,5,replace=False)
... 
array([9, 7, 4, 0, 6])
array([0, 9, 2, 4, 6])
array([2, 6, 5, 0, 9])
array([0, 3, 5, 7, 9])
array([0, 5, 9, 6, 7])
array([5, 0, 9, 6, 3])
array([7, 2, 6, 9, 1])
array([7, 6, 5, 8, 4])
array([6, 8, 5, 7, 4])
array([0, 1, 2, 3, 5])


One can see than an element cannot be selected more than one time.

### Weighted random sampling

To do weighted random sampling, it is possible to define for each element the probability to be selected:

>>> p = [0.05, 0.05, 0.1, 0.125, 0.175, 0.175, 0.125, 0.1, 0.05, 0.05]


Note: the sum must be equal to 1:

>>> sum(p)
1.0


Here for example the elements 0,1,8 or 9 will have a lower probability to be selected:

>>> for idx,p in enumerate(p):
...     print(p,data[idx])
... 
0.05 0
0.05 1
0.1 2
0.125 3
0.175 4
0.175 5
0.125 6
0.1 7
0.05 8
0.05 9


Lets check:

>>> for i in range(10):
...     np.random.choice(data,5,replace=False,p=p)
... 
array([7, 5, 0, 2, 3])
array([9, 2, 3, 5, 7])
array([2, 5, 3, 7, 4])
array([7, 2, 9, 4, 5])
array([1, 4, 6, 3, 2])
array([4, 5, 3, 7, 1])
array([2, 7, 4, 6, 3])
array([6, 5, 0, 1, 8])
array([4, 0, 5, 9, 6])
array([8, 9, 3, 4, 6])


### Random sampling for a 2D array

Lets consider the following 2D array:

>>> data =  np.arange(80).reshape((8, 10))
>>> data
array([[ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
       [20, 21, 22, 23, 24, 25, 26, 27, 28, 29],
       [30, 31, 32, 33, 34, 35, 36, 37, 38, 39],
       [40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
       [50, 51, 52, 53, 54, 55, 56, 57, 58, 59],
       [60, 61, 62, 63, 64, 65, 66, 67, 68, 69],
       [70, 71, 72, 73, 74, 75, 76, 77, 78, 79]])


data = \left( \begin{array}{ccc}
0 & 1 & 2 & 3 & 4 & 5 & 6 & 7 & 8 & 9 \\
10 & 11 & 12 & 13 & 14 & 15 & 16 & 17 & 18 & 19 \\
20 & 21 & 22 & 23 & 24 & 25 & 26 & 27 & 28 & 29 \\
30 & 31 & 32 & 33 & 34 & 35 & 36 & 37 & 38 & 39 \\
40 & 41 & 42 & 43 & 44 & 45 & 46 & 47 & 48 & 49 \\
50 & 51 & 52 & 53 & 54 & 55 & 56 & 57 & 58 & 59 \\
60 & 61 & 62 & 63 & 64 & 65 & 66 & 67 & 68 & 69 \\
70 & 71 & 72 & 73 & 74 & 75 & 76 & 77 & 78 & 79
\end{array}\right)

The function choice() takes only 1D array as an input, however a solution is to use ravel() to transform the 2D array to a 1D array, example:

>>> np.random.choice( data.ravel(),10,replace=False)
array([64, 35, 53, 14, 48, 29, 74, 21, 62, 41])