Pick Randomly From a Set of Options
Problem to Solve
I want to randomly choose one item from a collection of possible choices.
Sometimes, instead of generating a random integer from some range, you want to randomly pick one item from a collection of possible choices. Picking from a list or tuple can be done by generating a random integer and using it as an index, but that's unnecessary. It also doesn't work for a non-indexable collection like a set or dictionary.
The random module's choice function does this for you.
If you want to choose more than one item, and/or specify different weights (probabilities) for different items, use the choices function instead.
Recipes: Single Choice
To pick one item from a list, tuple, or range(...), or to pick one character from a string:
# Assuming the variable `S` already refers to an appropriate list/tuple/etc.:
random_element = random.choice(S)
To pick one item from a set, convert it to a list or tuple:
# Assuming the variable `some_set` already refers to an appropriate set:
random_element = random.choice(tuple(some_set))
To pick one item from a dictionary, use the keys() method to get a list of the keys and choose one of those randomly:
# Assuming the variable `some_dict` already refers to an appropriate dictionary:
random_key = random.choice(list(some_dict.keys()))
random_value = some_dict[random_key]
Recipes: Multiple Choices
The choices function can get a list of multiple items all randomly and independently chosen from a given sequence. The items are chosen "with replacement", which means an element is not "used up" when picked and can be picked again. As a result, it's entirely possible to choose more items than the length of the source sequence.
The first argument to choices is the sequence to choose from. If you want multiple items, pass a named argument k with the number of items you want to choose.
To get a sequence simulating ten consecutive rolls of a six-sided die:
roll_sequence = random.choices(range(1, 7), k=10)
roll_sum = sum(roll_sequence)
To get a string of twelve random vowels:
vowel_string = ''.join(random.choices('aeiou', k=12))
Side note: ''.join(list_of_strings_or_characters) is a weird-looking python idiom for concatenating multiple strings or characters into one string. The idea is that the string before the period is inserted between each of the elements of list_of_strings_or_characters. So, ', '.join(["one", "two", "three"]) would produce the string "one, two, three", and ''.join(["one", "two", "three"]) would produce the string "onetwothree" because the empty string '' is used to as the separator.
Recipes: Weighted Choices
Sometimes, you'll want to pick one item at random from a list of possibilities, but with some items being more likely than others. In that case, you can supply an optional second argument named weights to choices. It must have the same length as the sequence of options to choose from, and each element of weights should be a number that specifies the relative probability of the corresponding item in the sequence being chosen.
The weights don't have to be normalized, meaning that they don't have to sum to 1. The choices function will do that for you.
Let's say that (for some unfathomable reason) you want to choose three integers between 1 and 10 inclusive, but with a probability inversely proportional to the square of the integer. In other words, the number 1 should be one hundred times more likely than the number 10.
option_values = list(range(1, 11))
weight_values = [1 / (x ** 2) for x in option_values]
three_random_choices = random.choices(option_values, weights=weight_values, k=3)
If you only want one value chosen with a weighted probability, leave k unspecified so that it defaults to 1. However, you'll still get a list with one element in it, so you'll need to slap a bit of indexing on the end to extract that value from the list:
one_random_choice = random.choices(option_values, weights=weight_values)[0]
Unless you need unequal weighting, this makes random.choice more convenient than random.choices for picking a single item. For multiple items or unequal weighting, however, random.choices is the way to go.