List comprehensions are Python's most powerful syntactic feature — and the one most developers underuse. They let you build a new list from any iterable in a single, readable expression. Mastering them doesn't just make your code shorter; it makes it faster (the C-level bytecode is more optimized than an equivalent for loop) and more Pythonic.

Here are 10 patterns, ranging from beginner to production-grade, with before-and-after code so you can see exactly what each one replaces.

Pattern 1: Basic transformation

Python — square every number
# Before
squares = []
for n in range(10):
    squares.append(n ** 2)

# After — list comprehension
squares = [n ** 2 for n in range(10)]
# [0, 1, 4, 9, 16, 25, 36, 49, 64, 81]

Pattern 2: Filter with condition

Python — keep only even numbers
evens = [n for n in range(20) if n % 2 == 0]
# [0, 2, 4, 6, 8, 10, 12, 14, 16, 18]

# Filter strings: names longer than 4 characters
names = ["Ali", "Raman", "Bob", "Priya", "Sam"]
long_names = [n for n in names if len(n) > 4]
# ['Raman', 'Priya']

Pattern 3: Transform and filter together

Python
# Email addresses for users who have an @ sign and domain
raw_emails = ["raman@codewithraman.com", "invalid", "bob@example.com", ""]
valid = [e.lower().strip() for e in raw_emails if "@" in e and "." in e]
# ['raman@codewithraman.com', 'bob@example.com']

Pattern 4: Conditional expression (ternary)

Python — classify each value
scores = [45, 78, 92, 33, 67, 88]
grades = ["Pass" if s >= 50 else "Fail" for s in scores]
# ['Fail', 'Pass', 'Pass', 'Fail', 'Pass', 'Pass']

Pattern 5: Flatten a nested list

Python
matrix = [[1, 2, 3], [4, 5, 6], [7, 8, 9]]
flat = [n for row in matrix for n in row]
# [1, 2, 3, 4, 5, 6, 7, 8, 9]

# Order matters: outer loop first, inner loop second
# Same as: for row in matrix: for n in row: flat.append(n)

Pattern 6: Extract from dicts

Python — common data pipeline pattern
users = [
    {"name": "Raman", "age": 28, "active": True},
    {"name": "Alice", "age": 34, "active": False},
    {"name": "Bob",   "age": 22, "active": True},
]

# Get names of active users only
active_names = [u["name"] for u in users if u["active"]]
# ['Raman', 'Bob']

# Build a lookup dict — dict comprehension
name_to_age = {u["name"]: u["age"] for u in users}
# {'Raman': 28, 'Alice': 34, 'Bob': 22}

Pattern 7: String manipulation

Python
csv_line = " id , name , email , city "
headers = [h.strip().lower() for h in csv_line.split(",")]
# ['id', 'name', 'email', 'city']

# Generate slugs from titles
titles = ["SQL Window Functions", "Python for Data Engineers", "ERWIN Tutorial"]
slugs = [t.lower().replace(" ", "-") for t in titles]
# ['sql-window-functions', 'python-for-data-engineers', 'erwin-tutorial']

Pattern 8: Zip two lists

Python
columns = ["customer_id", "first_name", "email"]
values  = [101, "Raman", "raman@example.com"]

# Build INSERT VALUES clause
pairs = [f"{c} = {repr(v)}" for c, v in zip(columns, values)]
# ["customer_id = 101", "first_name = 'Raman'", "email = 'raman@example.com'"]

Pattern 9: Set and frozenset comprehensions

Python — deduplicate while transforming
tags = ["SQL", "sql", "Python", "PYTHON", "Java", "java"]
unique_lower = {t.lower() for t in tags}
# {'sql', 'python', 'java'} — set, unordered

Pattern 10: Generator expression (memory efficient)

Python — when you don't need the list in memory
# List comprehension — builds entire list in RAM
total = sum([n ** 2 for n in range(10_000_000)])

# Generator — computes on the fly, O(1) memory
total = sum(n ** 2 for n in range(10_000_000))  # just drop the []

When NOT to use list comprehensions

READABILITY FIRST If a comprehension needs more than one line to read, or has three or more nested loops, use a regular for loop. A comprehension that nobody can understand at a glance is worse than a slightly longer but clear for loop.
PRACTICE The Python Lists tutorial and Python Playground are the best places to practice these patterns. Try rewriting any for loop you see as a comprehension first, then decide which reads better.