[ Python 常見問題 ] Choose element(s) from List with different probability in Python

 Source From Here

Question
Have you ever wondered how to select random elements from a list with different probability in Python? In this article, we will discuss how to do the same. Let’s first consider the below example.

view plaincopy to clipboardprint?

1. import random  
2.     
3. sam_Lst = [10, 20, 3, 4, 100]  
4. ran = random.choice(sam_Lst)  
5. print(ran)  

In the above example, the probability of getting any element from the list is equal. But we want such methods in which the probability of choosing one element from the list is different. This is known as the weighted random choice in Python. In order to find weighted random choices in Python, there exist two ways:

* Relative weights
* Cumulative weights

HowTo
The function which will help us in this situation is random.choices(). This function allows making weighted random choices in python with replacement:

view plaincopy to clipboardprint?

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

Here, the ‘weight’ parameter plays an important role.

Case 1: Using Relative weights
The weight assigned to an element is known as relative weight.

Example1:

view plaincopy to clipboardprint?

1. import random  
2.     
3. # Creating a number list  
4. num_lst = [1, 22, 43, 19, 13, 29]  
5.     
6. print(random.choices(num_lst, weights=(14, 25, 30, 45, 55, 10), k=6))  

Output:

[19, 19, 13, 22, 13, 13]

In the above example, we assign weights to every element of the list. The weight of the element ‘13′ is highest i.e 55, so the probability of its occurrence is maximum. As we can see in the output, element 13 occurs 3 times, 19 occurs 2 times, and so on. So, now the probability of choosing an element from the list is different.

Example 2:

view plaincopy to clipboardprint?

1. import random  
2.     
3. # Creating a name list  
4. name_lst = ['October', 'November', 'December', 'January', 'March', 'June']  
5.     
6. print(random.choices(name_lst, weights=(40, 25, 30, 5, 15, 80), k=3))  

Output:

['June', 'October', 'June']

In the above example, the weight of element ‘June’ is maximum so its probability of selection will be maximum. And here, k=3 which means we are choosing only the top 3 elements from the list.

Case 2: Using Cumulative weights
The cumulative weight of an element is determined by adding the weight of its previous element and its own weight.

Example 1:

view plaincopy to clipboardprint?

1. import random  
2.     
3. # Creating a number list  
4. num_lst = [1, 22, 93, 19, 13, 25]  
5.     
6. print(random.choices(num_lst, cum_weights=(7, 13, 15, 20, 25, 20), k=6))  

Output:

[1, 22, 93, 22, 19, 1]

In the above example, the cumulative weight of element ‘19′ is maximum, so the probability of its selection will also be maximum.