Data Structures in Python

12 essential data structures for Python developers. Each one explained with Big O complexity, animated visuals, and real code samples you can copy.

Array

list

Fixed-size, contiguous block of memory. Elements are stored sequentially and accessed by index in constant time. The foundation of most other data structures.

7
3
9
1
5
[0]
[1]
[2]
[3]
[4]

arr[0] = 7

Complexity

Access by indexO(1)
SearchO(n)
Insert / DeleteO(n)

When to use

  • +You know the exact size at creation time
  • +You need the fastest possible index-based access
  • +Working with fixed-length data like matrices or buffers
Python
import array

# Fixed-type array (like C arrays)
nums = array.array("i", [10, 20, 30, 40, 50])

nums[0] = 99              # O(1) - direct index access
val = nums[2]             # O(1) - read by index

nums.append(60)           # O(1) amortized
nums.insert(1, 15)        # O(n) - shifts elements right

idx = nums.index(30)      # O(n) - linear search
nums.remove(15)           # O(n) - shifts elements left

sorted_nums = sorted(nums)  # O(n log n)

Dynamic Array

list

Resizable array that automatically grows when capacity is exceeded. The most commonly used data structure in most languages. Doubles its internal storage when full, giving amortized O(1) appends.

7
3
9
1
5
[0]
[1]
[2]
[3]
[4]

arr[0] = 7

Complexity

Access by indexO(1)
SearchO(n)
AppendO(1)*
Insert / Remove (middle)O(n)

When to use

  • +You need a resizable collection (most common case)
  • +You frequently access elements by index
  • +You mostly add or remove at the end
Python
names = ["Alice", "Bob", "Charlie"]

names.append("Diana")          # O(1) amortized
names.insert(1, "Eve")         # O(n) - shifts elements
names.pop()                    # O(1) - remove last
names.pop(0)                   # O(n) - shifts elements

has = "Bob" in names           # O(n) - linear scan
idx = names.index("Bob")       # O(n) - first occurrence

names.sort()                   # O(n log n) - Timsort
squares = [x * x for x in range(10)]  # O(n) list comprehension

Stack

list (as stack)

Last-In-First-Out (LIFO) collection. Only the top element is accessible. Used for tracking state that must be unwound in reverse order.

arrow_downward top
10

Push 10

Complexity

PushO(1)
PopO(1)
PeekO(1)
SearchO(n)

When to use

  • +Undo/redo functionality
  • +Expression parsing and evaluation
  • +DFS traversal of trees and graphs
  • +Matching brackets, parentheses validation
Python
stack: list[int] = []

stack.append(10)        # O(1) - push
stack.append(20)
stack.append(30)

top = stack[-1]         # O(1) - peek -> 30
val = stack.pop()       # O(1) - pop -> 30

# Interview pattern: valid parentheses
def is_valid(s: str) -> bool:
    pairs = {"(": ")", "[": "]", "{": "}"}
    st: list[str] = []
    for c in s:
        if c in pairs:
            st.append(pairs[c])
        elif not st or st.pop() != c:
            return False
    return len(st) == 0

Queue

collections.deque

First-In-First-Out (FIFO) collection. Elements are added at the back and removed from the front. Fundamental for breadth-first processing.

front
10
back

Enqueue 10

Complexity

EnqueueO(1)
DequeueO(1)
PeekO(1)
SearchO(n)

When to use

  • +BFS traversal of trees and graphs
  • +Task scheduling and job queues
  • +Message passing between components
  • +Rate limiting, buffering
Python
from collections import deque

queue: deque[str] = deque()

queue.append("Task A")      # O(1) - enqueue
queue.append("Task B")
queue.append("Task C")

first = queue[0]            # O(1) - peek front
val = queue.popleft()       # O(1) - dequeue -> "Task A"

# Interview pattern: BFS
def bfs(root):
    q = deque([root])
    while q:
        node = q.popleft()           # O(1)
        print(node.val, end=" ")
        if node.left:  q.append(node.left)
        if node.right: q.append(node.right)

Hash Map

dict

Maps keys to values using a hash function for near-constant-time lookups. The single most important data structure for coding interviews. Every language has a built-in implementation.

0
age:30
1
2
name:Al
3
city:NY

hash("age") = 0

Complexity

InsertO(1)*
LookupO(1)
DeleteO(1)
Contains keyO(1)

When to use

  • +Two Sum and frequency counting patterns
  • +Caching computed results (memoization)
  • +Grouping data by a key
  • +Any problem requiring O(1) lookup by key
Python
prices = {"apple": 3, "banana": 5}

prices["cherry"] = 2              # O(1) - add
prices["apple"] = 10              # O(1) - update
has = "banana" in prices          # O(1) - key check
del prices["cherry"]              # O(1) - remove

count = prices.get("mango", 0)   # O(1) - safe lookup

# Interview pattern: Two Sum
def two_sum(nums: list[int], target: int) -> list[int]:
    seen: dict[int, int] = {}
    for i, num in enumerate(nums):
        need = target - num
        if need in seen:            # O(1) lookup
            return [seen[need], i]
        seen[num] = i               # O(1) insert
    return []

Hash Set

set

Unordered collection of unique elements. Uses hashing internally for O(1) membership testing. Supports mathematical set operations like union, intersection, and difference.

0
age:30
1
2
name:Al
3
city:NY

hash("age") = 0

Complexity

AddO(1)*
ContainsO(1)
RemoveO(1)
Union / IntersectO(n)

When to use

  • +Checking if an element exists in O(1)
  • +Removing duplicates from a collection
  • +Set operations: union, intersection, difference
  • +Visited tracking in graph traversal
Python
s = {1, 2, 3, 4, 5}

s.add(6)                # O(1)
s.discard(1)            # O(1) - no error if missing
has = 4 in s            # O(1) -> True

# Set operations
other = {4, 5, 6, 7}
s & other               # intersection -> {4, 5, 6}
s | other               # union
s - other               # difference

# Interview pattern: contains duplicate
def contains_duplicate(nums: list[int]) -> bool:
    return len(nums) != len(set(nums))

Linked List

collections.deque

Sequence of nodes where each node points to the next (singly linked) or both next and previous (doubly linked). Efficient insertion and deletion at any known position, but no index-based access.

5
12
8
20

traversing: 5

Complexity

Insert at head/tailO(1)
Remove (given node)O(1)
SearchO(n)
Access by indexO(n)

When to use

  • +Frequent insertion/deletion in the middle
  • +Implementing LRU cache (with a hash map)
  • +When you need a deque (double-ended queue)
  • +Problems involving pointer manipulation
Python
from collections import deque

# deque is a doubly-linked list under the hood
dll: deque[int] = deque()

dll.append(10)          # O(1) - add right
dll.appendleft(5)       # O(1) - add left
dll.pop()               # O(1) - remove right
dll.popleft()           # O(1) - remove left

# Interview pattern: reverse a linked list
class ListNode:
    def __init__(self, val=0, next=None):
        self.val, self.next = val, next

def reverse(head: ListNode | None) -> ListNode | None:
    prev, curr = None, head
    while curr:
        curr.next, prev, curr = prev, curr, curr.next
    return prev

Sorted Set (BST)

SortedList (sortedcontainers)

Collection of unique elements maintained in sorted order, typically backed by a balanced binary search tree (red-black tree). Supports range queries and O(log n) min/max.

831215

search(8)

Complexity

AddO(log n)
ContainsO(log n)
RemoveO(log n)
Min / MaxO(log n)

When to use

  • +Maintaining a sorted collection of unique items
  • +Range queries (all elements between X and Y)
  • +Sliding window problems needing sorted order
  • +Leaderboards, ranking systems
Python
# pip install sortedcontainers
from sortedcontainers import SortedList

sl = SortedList([5, 3, 8, 1, 9])
# Maintains sorted order: [1, 3, 5, 8, 9]

sl.add(4)                    # O(log n)
sl.discard(3)                # O(log n)

minimum = sl[0]              # O(1) -> 1
maximum = sl[-1]             # O(1) -> 9

# Range query
left = sl.bisect_left(4)    # O(log n)
right = sl.bisect_right(8)  # O(log n)
between = list(sl[left:right])  # [4, 5, 8]

Sorted Map (BST)

SortedDict (sortedcontainers)

Key-value pairs maintained in sorted key order, typically backed by a balanced BST. Enables ordered iteration and range lookups that hash maps cannot provide.

831215

search(8)

Complexity

InsertO(log n)
LookupO(log n)
RemoveO(log n)
Iterate (sorted)O(n)

When to use

  • +You need sorted key-value pairs
  • +Ordered iteration over entries
  • +Range lookups by key
  • +When insertion order or sorted order matters
Python
# pip install sortedcontainers
from sortedcontainers import SortedDict

sd = SortedDict({"banana": 2, "apple": 5, "cherry": 1})

sd["date"] = 3              # O(log n)
val = sd["apple"]           # O(log n) -> 5
del sd["cherry"]            # O(log n)

# Iterates in sorted key order
for key, value in sd.items():
    print(f"{key}: {value}")
# apple: 5, banana: 2, date: 3

first_key = sd.keys()[0]    # O(1) -> "apple"
last_key = sd.keys()[-1]    # O(1) -> "date"

Priority Queue (Heap)

heapq

Collection where elements are dequeued by priority rather than insertion order. Typically implemented as a binary heap. Essential for shortest-path algorithms and top-K problems.

13579

min-heap

Complexity

InsertO(log n)
Extract min/maxO(log n)
PeekO(1)
SearchO(n)

When to use

  • +Dijkstra's shortest path algorithm
  • +Merge K sorted lists/streams
  • +Top-K / Kth largest element problems
  • +Event-driven simulation, scheduling
Python
import heapq

heap: list[int] = []

heapq.heappush(heap, 3)       # O(log n)
heapq.heappush(heap, 1)
heapq.heappush(heap, 2)

top = heap[0]                  # O(1) - peek min -> 1
val = heapq.heappop(heap)     # O(log n) - pop min -> 1

# Max-heap: negate values
max_heap: list[int] = []
heapq.heappush(max_heap, -10)
largest = -heapq.heappop(max_heap)  # -> 10

# Interview pattern: K closest points
def k_closest(points, k):
    heap = [(x*x + y*y, [x, y]) for x, y in points]
    heapq.heapify(heap)            # O(n)
    return [heapq.heappop(heap)[1] for _ in range(k)]

Concurrent Hash Map

threading.Lock + dict

Thread-safe hash map designed for concurrent read/write access from multiple threads. Uses fine-grained locking or lock-free techniques instead of a single global lock.

0
age:30
1
2
name:Al
3
city:NY

hash("age") = 0

Complexity

InsertO(1)*
LookupO(1)
DeleteO(1)
Atomic updateO(1)*

When to use

  • +Multi-threaded caching
  • +Shared state across threads or async tasks
  • +Producer-consumer patterns with keyed data
  • +When you need concurrent reads and writes
Python
import threading

class ThreadSafeDict:
    def __init__(self):
        self._data: dict = {}
        self._lock = threading.Lock()

    def get(self, key, default=None):
        with self._lock:
            return self._data.get(key, default)

    def set(self, key, value):
        with self._lock:
            self._data[key] = value

    def get_or_add(self, key, factory):
        with self._lock:
            if key not in self._data:
                self._data[key] = factory(key)
            return self._data[key]

cache = ThreadSafeDict()
cache.set("counter", 0)

Memory View / Slice

memoryview

Zero-copy view over a contiguous region of memory. Lets you reference a portion of an array or buffer without allocating new memory. Critical for performance-sensitive parsing and processing.

1
2
3
4
5
6

Span[0..3] = [1, 2, 3]

Complexity

Create sliceO(1)
Access by indexO(1)
SearchO(n)
CopyO(n)

When to use

  • +Parsing strings or binary data without copies
  • +Processing sub-arrays without allocation
  • +High-performance, zero-allocation code paths
  • +Interop with native or unmanaged memory
Python
# memoryview - zero-copy view over bytes/bytearray
data = bytearray(b"Hello, World!")
view = memoryview(data)

# Zero-copy slicing
chunk = view[7:12]            # O(1) - view of "World"
chunk[0] = ord("E")          # mutates original
print(data)                   # bytearray(b"Hello, Eorld!")

# Efficient binary parsing
header = view[:5]             # O(1) - no copy
body = view[7:]               # O(1) - no copy

view.release()                # free the buffer lock

Big O Comparison

Average-case time complexity. * = amortized.

StructureAccessSearchInsertDelete
ArrayO(1)O(n)O(n)O(n)
Dynamic ArrayO(1)O(n)O(1)*O(n)
StackO(n)O(n)O(1)*O(1)
QueueO(n)O(n)O(1)*O(1)
Hash MapO(1)O(1)O(1)*O(1)
Hash SetN/AO(1)O(1)*O(1)
Linked ListO(n)O(n)O(1)O(1)
Sorted SetO(n)O(log n)O(log n)O(log n)
Sorted MapO(log n)O(log n)O(log n)O(log n)
Priority QueueO(n)O(n)O(log n)O(log n)
Concurrent MapO(1)O(1)O(1)*O(1)
Memory ViewO(1)O(n)N/AN/A

Which collection should I use?

I need to...Use
Store items by index, resize dynamicallyList / Dynamic Array
Map keys to values with O(1) lookupHashMap / Dictionary
Track unique items, check existence in O(1)HashSet / Set
Last-in-first-out (undo, DFS, brackets)Stack
First-in-first-out (BFS, task queues)Queue
Keep elements sorted at all timesSortedSet / TreeSet
Process items by priority (Dijkstra, top-K)PriorityQueue / Heap
Insert/delete at a known position in O(1)LinkedList
Sorted key-value pairsSortedDictionary / TreeMap
Thread-safe shared cacheConcurrentDictionary
Slice arrays/strings without copyingSpan / Slice / memoryview

Frequently Asked Questions

What are the most important data structures in Python?add

The most commonly used are dynamic arrays (List/ArrayList/vector), hash maps (Dictionary/HashMap/dict), and hash sets. For interviews, also know stacks, queues, trees, and priority queues. These cover 90%+ of coding interview problems.

Which Python data structure should I learn first?add

Start with the dynamic array and hash map. Together they solve the majority of interview problems. Then learn stacks (for DFS, bracket matching) and queues (for BFS). After that, tackle trees, heaps, and graphs.

Does Big O complexity change between languages?add

No. Big O measures algorithmic complexity, not language-specific performance. A hash map lookup is O(1) whether you use Python dict, Java HashMap, or C# Dictionary. Constant factors differ (C++ is faster than Python in wall-clock time), but Big O is the same.

Is there a built-in priority queue in Python?add

Yes. The heapq module provides a min-heap. Push with heapq.heappush() and pop with heapq.heappop(). For a max-heap, negate the values.

Know the structures.
Now use them under pressure.

Practice with an AI interviewer that asks you to implement, optimize, and explain data structure choices in real time.

Try a free mock interviewarrow_forward