How to Get Weighted Random Choice in Python?

Python's weighted random choice allows you to select items from a list where each item has a different probability of being chosen. Unlike simple random selection where each item has equal chances, weighted selection lets you control the likelihood of each item being picked.

Syntax

The primary method for weighted random choice in Python is random.choices():

random.choices(population, weights=None, cum_weights=None, k=1)

Parameters:

• population The list of items to choose from (required)

• weights List of weights corresponding to each item (optional)

• cum_weights List of cumulative weights (optional)

• k Number of items to select (default is 1)

Using random.choices()

The most straightforward approach uses Python's built-in random.choices() function with explicit weights:

import random

colors = ['Red', 'Blue', 'Green']
weights = [0.6, 0.3, 0.1]  # 60%, 30%, 10% probability

chosen = random.choices(colors, weights, k=5)
print(chosen)

['Red', 'Red', 'Blue', 'Red', 'Red']

In this example, 'Red' has the highest weight (0.6), so it appears more frequently in the results.

Using numpy.random.choice()

NumPy provides an alternative approach with numpy.random.choice():

import numpy as np

colors = ['Red', 'Blue', 'Green']
weights = [0.6, 0.3, 0.1]

chosen = np.random.choice(colors, size=5, p=weights)
print(chosen)

['Red' 'Red' 'Green' 'Red' 'Blue']

Note that NumPy uses the parameter p for probabilities and size instead of k.

Using Cumulative Weights

Instead of individual weights, you can use cumulative weights:

import random

items = ['A', 'B', 'C', 'D']
cum_weights = [10, 30, 60, 100]  # Cumulative weights

result = random.choices(items, cum_weights=cum_weights, k=8)
print(result)

['D', 'C', 'D', 'B', 'D', 'C', 'D', 'C']

Comparison

Conclusion

Python offers multiple ways to perform weighted random selection. Use random.choices() for simple cases with built-in functionality, or numpy.random.choice() when working with NumPy arrays and scientific computing.