Sometimes data we store or retrieve in an application can have little or no order. We may have to rearrange the data to correctly process it or efficiently use it. Over the years, computer scientists have created many sorting algorithms to organize data.
In this article we'll have a look at popular sorting algorithms, understand how they work and code them in Python. We'll also compare how quickly they sort items in a list.
For simplicity, algorithm implementations would be sorting lists of numbers in ascending order. Of course, you're free to adapt them to your needs
If you'd like to learn about a specific algorithm, you can jump to it here:
This simple sorting algorithm iterates over a list, comparing elements in pairs and swapping them until the larger elements "bubble up" to the end of the list, and the smaller elements stay at the "bottom".
We begin by comparing the first two elements of the list. If the first element is larger than the second element, we swap them. If they are already in order we leave them as is. We then move to the next pair of elements, compare their values and swap as necessary. This process continues to the last pair of items in the list.
Upon reaching the end of the list, it repeats this process for every item. Though, this is highly inefficient. What if only a single swap needs to be made in the array? Why would we still iterate though it n^2 times, even though it's already sorted?
Obviously, to optimize the algorithm, we need to stop it when it's finished sorting.
How would we know that we're finished sorting? If the items were in order then we would not have to swap items. So, whenever we swap values we set a flag to
True to repeat sorting process. If no swaps occurred, the flag would remain
False and the algorithm would stop.
With the optimization, we can implement the bubble sort in Python as follows:
def bubble_sort(nums): # We set swapped to True so the loop looks runs at least once swapped = True while swapped: swapped = False for i in range(len(nums) - 1): if nums[i] > nums[i + 1]: # Swap the elements nums[i], nums[i + 1] = nums[i + 1], nums[i] # Set the flag to True so we'll loop again swapped = True # Verify it works random_list_of_nums = [5, 2, 1, 8, 4] bubble_sort(random_list_of_nums) print(random_list_of_nums)
The algorithm runs in a
while loop, only breaking when no items are swapped. We set
True in the beginning to ensure that the algorithm runs at least once.
In the worst case scenario (when the list is in reverse order), this algorithm would have to swap every single item of the array. Our
swapped flag would be set to
True on every iteration. Therefore, if we have n elements in our list, we would have n iterations per item - thus Bubble Sort's time complexity is O(n^2).
This algorithm segments the list into two parts: sorted and unsorted. We continuously remove the smallest element of the unsorted segment of the list and append it to the sorted segment.
In practice, we don't need to create a new list for the sorted elements, what we do is treat the leftmost part of the list as the sorted segment. We then search the entire list for the smallest element, and swap it with the first element.
Now we know that the first element of the list is sorted, we get the smallest element of the remaining items and swap it with the second element. This iterates until the last item of the list is remaining element to be examined.
def selection_sort(nums): # This value of i corresponds to how many values were sorted for i in range(len(nums)): # We assume that the first item of the unsorted segment is the smallest lowest_value_index = i # This loop iterates over the unsorted items for j in range(i + 1, len(nums)): if nums[j] < nums[lowest_value_index]: lowest_value_index = j # Swap values of the lowest unsorted element with the first unsorted # element nums[i], nums[lowest_value_index] = nums[lowest_value_index], nums[i] # Verify it works random_list_of_nums = [12, 8, 3, 20, 11] selection_sort(random_list_of_nums) print(random_list_of_nums)
We see that as
i increases, we need to need to check less items.
We can easily get the time complexity by examining the
for loops in the Selection Sort algorithm. For a list with n elements, the outer loop iterates n times. The inner loop iterate n-1 when i is equal to 1, and then n-2 as i is equal to 2 and so forth.
The amount of comparisons are
(n - 1) + (n - 2) + ... + 1, which gives the Selection Sort a time complexity of O(n^2).
Like Selection Sort, this algorithm segments the list into sorted and unsorted parts. It iterates over the unsorted segment, and inserts the element being viewed into the correct position of the sorted list.
We assume that the first element of the list is sorted. We then go to the next element, let's call it
x is larger than the first element we leave as is. If
x is smaller, we copy the value of the first element to the second position and then set the first element to
As we go to the other elements of the unsorted segment, we continuously move larger elements in the sorted segment up the list until we encounter an element smaller than
x or reach the end of the sorted segment, and then place
x in it's correct position.
def insertion_sort(nums): # Start on the second element as we assume the first element is sorted for i in range(1, len(nums)): item_to_insert = nums[i] # And keep a reference of the index of the previous element j = i - 1 # Move all items of the sorted segment forward if they are larger than # the item to insert while j >= 0 and nums[j] > item_to_insert: nums[j + 1] = nums[j] j -= 1 # Insert the item nums[j + 1] = item_to_insert # Verify it works random_list_of_nums = [9, 1, 15, 28, 6] insertion_sort(random_list_of_nums) print(random_list_of_nums)
In the worst case scenario, an array would be sorted in reverse order. The outer
for loop in the Insertion Sort function always iterates n-1 times.
In the worst case scenario, the inner for loop would swap once, then swap two and so forth. The amount of swaps would then be
1 + 2 + ... + (n - 3) + (n - 2) + (n - 1) which gives the Insertion Sort a time complexity of O(n^2).
This popular sorting algorithm, like the Insertion and Selection sorts, segments the list into sorted and unsorted parts. It converts the unsorted segment of the list to a Heap data structure, so that we can efficiently determine the largest element.
We begin by transforming the list into a Max Heap - a Binary Tree where the biggest element is the root node. We then place that item to the end of the list. We then rebuild our Max Heap which now has one less value, placing the new largest value before the last item of the list.
We iterate this process of building the heap until all nodes are removed.
We create an helper function
heapify to implement this algorithm:
def heapify(nums, heap_size, root_index): # Assume the index of the largest element is the root index largest = root_index left_child = (2 * root_index) + 1 right_child = (2 * root_index) + 2 # If the left child of the root is a valid index, and the element is greater # than the current largest element, then update the largest element if left_child < heap_size and nums[left_child] > nums[largest]: largest = left_child # Do the same for the right child of the root if right_child < heap_size and nums[right_child] > nums[largest]: largest = right_child # If the largest element is no longer the root element, swap them if largest != root_index: nums[root_index], nums[largest] = nums[largest], nums[root_index] # Heapify the new root element to ensure it's the largest heapify(nums, heap_size, largest) def heap_sort(nums): n = len(nums) # Create a Max Heap from the list # The 2nd argument of range means we stop at the element before -1 i.e. # the first element of the list. # The 3rd argument of range means we iterate backwards, reducing the count # of i by 1 for i in range(n, -1, -1): heapify(nums, n, i) # Move the root of the max heap to the end of for i in range(n - 1, 0, -1): nums[i], nums = nums, nums[i] heapify(nums, i, 0) # Verify it works random_list_of_nums = [35, 12, 43, 8, 51] heap_sort(random_list_of_nums) print(random_list_of_nums)
Let's first look at the time complexity of the
heapify function. In the worst case the largest element is never the root element, this causes a recursive call to
heapify. While recursive calls might seem dauntingly expensive, remember that we're working with a binary tree.
Visualize a binary tree with 3 elements, it has a height of 2. Now visualize a binary tree with 7 elements, it has a height of 3. The tree grows logarithmically to n. The
heapify function traverses that tree in O(log(n)) time.
heap_sort function iterates over the array n times. Therefore the overall time complexity of the Heap Sort algorithm is O(nlog(n)).
This divide and conquer algorithm splits a list in half, and keeps splitting the list by 2 until it only has singular elements.
Adjacent elements become sorted pairs, then sorted pairs are merged and sorted with other pairs as well. This process continues until we get a sorted list with all the elements of the unsorted input list.
We recursively split the list in half until we have lists with size one. We then merge each half that was split, sorting them in the process.
Sorting is done by comparing the smallest elements of each half. The first element of each list are the first to be compared. If the first half begins with a smaller value, then we add that to the sorted list. We then compare the second smallest value of the first half with the first smallest value of the second half.
Every time we select the smaller value at the beginning of a half, we move the index of which item needs to be compared by one.
def merge(left_list, right_list): sorted_list =  left_list_index = right_list_index = 0 # We use the list lengths often, so its handy to make variables left_list_length, right_list_length = len(left_list), len(right_list) for _ in range(left_list_length + right_list_length): if left_list_index < left_list_length and right_list_index < right_list_length: # We check which value from the start of each list is smaller # If the item at the beginning of the left list is smaller, add it # to the sorted list if left_list[left_list_index] <= right_list[right_list_index]: sorted_list.append(left_list[left_list_index]) left_list_index += 1 # If the item at the beginning of the right list is smaller, add it # to the sorted list else: sorted_list.append(right_list[right_list_index]) right_list_index += 1 # If we've reached the end of the of the left list, add the elements # from the right list elif left_list_index == left_list_length: sorted_list.append(right_list[right_list_index]) right_list_index += 1 # If we've reached the end of the of the right list, add the elements # from the left list elif right_list_index == right_list_length: sorted_list.append(left_list[left_list_index]) left_list_index += 1 return sorted_list def merge_sort(nums): # If the list is a single element, return it if len(nums) <= 1: return nums # Use floor division to get midpoint, indices must be integers mid = len(nums) // 2 # Sort and merge each half left_list = merge_sort(nums[:mid]) right_list = merge_sort(nums[mid:]) # Merge the sorted lists into a new one return merge(left_list, right_list) # Verify it works random_list_of_nums = [120, 45, 68, 250, 176] random_list_of_nums = merge_sort(random_list_of_nums) print(random_list_of_nums)
Note that the
merge_sort() function, unlike the previous sorting algorithms, returns a new list that is sorted, rather than sorting the existing list.
Therefore, Merge Sort requires space to create a new list of the same size as the input list.
Let's first look at the
merge function. It takes two lists, and iterates n times, where n is the size of their combined input. The
merge_sort function splits its given array in 2, and recursively sorts the sub-arrays. As the input being recursed is half of what was given, like binary trees this makes the time it takes to process grow logarithmically to n.
Therefore the overall time complexity of the Merge Sort algorithm is O(nlog(n)).
This divide and conquer algorithm is the most often used sorting algorithm covered in this article. When configured correctly, it's extremely efficient and does not require the extra space Merge Sort uses. We partition the list around a pivot element, sorting values around the pivot.
Quick Sort begins by partitioning the list - picking one value of the list that will be in its sorted place. This value is called a pivot. All elements smaller than the pivot are moved to its left. All larger elements are moved to its right.
Knowing that the pivot is in it's rightful place, we recursively sort the values around the pivot until the entire list is sorted.
# There are different ways to do a Quick Sort partition, this implements the # Hoare partition scheme. Tony Hoare also created the Quick Sort algorithm. def partition(nums, low, high): # We select the middle element to be the pivot. Some implementations select # the first element or the last element. Sometimes the median value becomes # the pivot, or a random one. There are many more strategies that can be # chosen or created. pivot = nums[(low + high) // 2] i = low - 1 j = high + 1 while True: i += 1 while nums[i] < pivot: i += 1 j -= 1 while nums[j] > pivot: j -= 1 if i >= j: return j # If an element at i (on the left of the pivot) is larger than the # element at j (on right right of the pivot), then swap them nums[i], nums[j] = nums[j], nums[i] def quick_sort(nums): # Create a helper function that will be called recursively def _quick_sort(items, low, high): if low < high: # This is the index after the pivot, where our lists are split split_index = partition(items, low, high) _quick_sort(items, low, split_index) _quick_sort(items, split_index + 1, high) _quick_sort(nums, 0, len(nums) - 1) # Verify it works random_list_of_nums = [22, 5, 1, 18, 99] quick_sort(random_list_of_nums) print(random_list_of_nums)
The worst case scenario is when the smallest or largest element is always selected as the pivot. This would create partitions of size n-1, causing recursive calls n-1 times. This leads us to a worst case time complexity of O(n^2).
While this is a terrible worst case, Quick Sort is heavily used because it's average time complexity is much quicker. While the
partition function utilizes nested while loops, it does comparisons on all elements of the array to make its swaps. As such, it has a time complexity of O(n).
With a good pivot, the Quick Sort function would partition the array into halves which grows logarithmically with n. Therefore the average time complexity of the Quick Sort algorithm is O(nlog(n)).
Python's Built-in Sort Functions
While it's beneficial to understand these sorting algorithms, in most Python projects you would probably use the sort functions already provided in the language.
We can change our list to have it's contents sorted with the
apples_eaten_a_day = [2, 1, 1, 3, 1, 2, 2] apples_eaten_a_day.sort() print(apples_eaten_a_day) # [1, 1, 1, 2, 2, 2, 3]
Or we can use the
sorted() function to create a new sorted list:
apples_eaten_a_day_2 = [2, 1, 1, 3, 1, 2, 2] sorted_apples = sorted(apples_eaten_a_day_2) print(sorted_apples) # [1, 1, 1, 2, 2, 2, 3]
They both sort in ascending order, but you can easily sort in descending order by setting the
reverse flag to
# Reverse sort the list in-place apples_eaten_a_day.sort(reverse=True) print(apples_eaten_a_day) # [3, 2, 2, 2, 1, 1, 1] # Reverse sort to get a new list sorted_apples_desc = sorted(apples_eaten_a_day_2, reverse=True) print(sorted_apples_desc) # [3, 2, 2, 2, 1, 1, 1]
Unlike the sorting algorithm functions we created, both these functions can sort lists of tuples and classes. The
sorted() function can sort any iterable object, that includes - lists, strings, tuples, dictionaries, sets, and custom iterators you can create.
These sort functions implement the Tim Sort algorithm, an algorithm inspired by Merge Sort and Insertion Sort.
To get an idea of how quickly they perform, we generate a list of 5000 numbers between 0 and 1000. We then time how long it takes for each algorithm to complete. This is repeated 10 times so that we can more reliably establish a pattern of performance.
These were the results, the time is in seconds:
You would get different values if you set up the test yourself, but the patterns observed should be the same or similar. Bubble Sort is the slowest the worst performer of all the algorithms. While it useful as an introduction to sorting and algorithms, it's not fit for practical use.
We also notice that Quick Sort is very fast, nearly twice as fast as Merge Sort and it wouldn't need as much space to run. Recall that our partition was based on the middle element of the list, different partitions could have different outcomes.
As Insertion Sort performs much less comparisons than Selection Sort, the implementations are usually quicker but in these runs Selection Sort is slightly faster.
Insertion Sorts does much more swaps than Selection Sort. If swapping values takes up considerably more time than comparing values, then this "contrary" result would be plausible.
Be mindful of the environment when choosing your sorting algorithm, as it will affect performance.
Sorting algorithms gives us many ways to order our data. We looked at 6 different algorithms - Bubble Sort, Selection Sort, Insertion Sort, Merge Sort, Heap Sort, Quick Sort - and their implementations in Python.
The amount of comparison and swaps the algorithm performs along with the environment the code runs are key determinants of performance. In real Python applications, it's recommended we stick with the built in Python sort functions for their flexibility on the input and speed.