How to choose elements from the list with different probability using NumPy?

NumPy provides several methods to randomly choose elements from a list with different probabilities. The probability values must sum to 1.0. Here are three main approaches using NumPy's random module.

Using numpy.random.choice()

The choice() function randomly samples elements from a 1-D array with specified probabilities ?

Syntax

numpy.random.choice(a, size=None, replace=True, p=None)

Parameters:

• a Input array or list of elements

• size Output shape (optional)

• replace Whether sampling is with replacement (default: True)

• p Probabilities for each element (must sum to 1)

Example 1: 1-D Array

import numpy as np

elements = [10, 12, 4, 5, 98]
probabilities = [0.1, 0.2, 0.3, 0.2, 0.2]

arr = np.random.choice(elements, size=8, p=probabilities)
print("Random selection:", arr)

Random selection: [10  4  5 10 98 12  4 12]

Example 2: 2-D Array

import numpy as np

elements = [1, 0, 12, 4, 5, 98, 34]
probabilities = [0.1, 0.1, 0.2, 0.2, 0.2, 0.1, 0.1]

arr = np.random.choice(elements, size=(3, 2), p=probabilities)
print("2-D random selection:")
print(arr)

2-D random selection:
[[ 4  5]
 [12  4]
 [ 5 34]]

Using numpy.random.multinomial()

The multinomial() function generates samples from a multinomial distribution, returning counts of occurrences ?

Syntax

numpy.random.multinomial(n, pvals, size=None)

Parameters:

• n Number of trials

• pvals Probabilities for each outcome

• size Output shape (optional)

Example

import numpy as np

probabilities = [0.1, 0.2, 0.3, 0.2, 0.2]

# Single trial
result1 = np.random.multinomial(10, probabilities)
print("Single trial (10 samples):", result1)

# Multiple trials
result2 = np.random.multinomial(10, probabilities, size=3)
print("Multiple trials:")
print(result2)

Single trial (10 samples): [1 2 3 2 2]
Multiple trials:
[[0 1 4 3 2]
 [2 2 2 2 2]
 [1 1 3 3 2]]

Using numpy.random.default_rng().choice()

The modern approach using NumPy's new random number generator (available in NumPy 1.17+) ?

Example 1: 1-D Selection

import numpy as np

rng = np.random.default_rng(42)  # Seed for reproducibility
elements = [1, 2, 3, 4, 10]
probabilities = [0.1, 0.1, 0.2, 0.3, 0.3]

arr = rng.choice(elements, p=probabilities, size=10)
print("Modern RNG selection:", arr)

Modern RNG selection: [10  4 10  4  4  3 10  4  3  4]

Example 2: 2-D Selection

import numpy as np

rng = np.random.default_rng(42)
elements = [23, 43, 42, 5, 78, 90]
probabilities = [0.1, 0.2, 0.2, 0.3, 0.1, 0.1]

arr = rng.choice(elements, p=probabilities, size=(3, 3))
print("2-D selection with modern RNG:")
print(arr)

2-D selection with modern RNG:
[[78  5 43]
 [ 5  5 42]
 [43 42 90]]

Comparison

Conclusion

Use numpy.random.choice() for simple weighted sampling, multinomial() for count-based distributions, and default_rng().choice() for modern NumPy applications. All methods require probabilities that sum to 1.0.