lesson 63 Comprehensions · intermediate

Distinct values with a set comprehension

A set comprehension collects the distinct values a collection contains. Repeats disappear, and so does the order.

You'll be able to

Build a set of distinct values with {expression for item in items}
Compare two sets with difference and union to answer who-is-missing questions
Sort a set when you need to show it, because a set has no order and no positions

The idea

*Which countries do our users come from?* is not *list every user's country*: you want each country once. In the Collection Algorithms module you skipped values already seen. A set does that by itself: it holds each value at most once.

A set comprehension uses braces without a colon: {expression for item in items}. A set has no order, so when you print it for people, sort it first with sorted, which gives a list.

Run each example to see what it prints. Running examples doesn't count toward progress.

example-1.py · Countries our users come from

users = [
    {"user_id": 6078, "country": "Canada"},
    {"user_id": 7345, "country": "Brazil"},
    {"user_id": 7712, "country": "Canada"},
    {"user_id": 8020, "country": "India"},
]
countries = {user["country"] for user in users}
print(len(countries))
print(sorted(countries))

Because a set has no order, it has no positions either. Asking for the first item is an error, not the first country you added.

example-2.py · A set has no position 0

countries = {"Canada", "Brazil", "India"}
print(countries[0])

Sets earn their keep when you compare two collections. Build a set from each, then let difference (see Dictionaries & Sets) answer *who is in the first but not the second*. Checking in against a set is also fast however large it gets, which matters when you check thousands of IDs.

example-3.py · Users who never rated a film

def never_rated(users, ratings):
    everyone = {user["user_id"] for user in users}
    raters = {rating["user_id"] for rating in ratings}
    return everyone.difference(raters)

users = [{"user_id": 6078}, {"user_id": 7345}, {"user_id": 7712}, {"user_id": 8020}]
ratings = [
    {"user_id": 6078, "movie_id": 2439},
    {"user_id": 6078, "movie_id": 935},
    {"user_id": 8020, "movie_id": 266},
]
print(sorted(never_rated(users, ratings)))

When to use it: distinct values (countries, tags, airports), and *in A but not B* questions. Mistakes to avoid: expecting the set to keep the order you added things in; indexing a set; and writing {} for an empty set. Empty braces make an empty dictionary; an empty set is written set().

Review

{expression for item in items} keeps each value once. Normalise first if different spellings should count as one.
A set has no order and no positions: sort it to show it, and use difference or union to compare.
{} is an empty dictionary; an empty set is set().
Next: two for clauses in one comprehension, to flatten lists inside lists.

Practise this