np.argsort() for sort indices in NumPy - Step-by-Step Execution

Concept Flow - np.argsort() for sort indices

Input Array

↓

Compute Indices that Sort

↓

Return Sorted Indices Array

↓

Use Indices to Access Sorted Elements

np.argsort() takes an array and returns the indices that would sort the array. These indices can then be used to reorder the original array.

Execution Sample

NumPy

import numpy as np
arr = np.array([30, 10, 20])
indices = np.argsort(arr)
sorted_arr = arr[indices]

This code finds the indices that sort the array and then uses them to get the sorted array.

Execution Table

Step	Action	Array State	Indices Computed	Result
1	Start with array	[30, 10, 20]	N/A	N/A
2	Call np.argsort(arr)	[30, 10, 20]	[1, 2, 0]	Indices that sort array
3	Use indices to reorder array	[30, 10, 20]	[1, 2, 0]	[10, 20, 30]
4	End	[30, 10, 20]	[1, 2, 0]	Sorted array obtained

💡 All steps completed; indices correctly represent sorting order

Variable Tracker

Variable	Start	After np.argsort	After indexing	Final
arr	[30, 10, 20]	[30, 10, 20]	[30, 10, 20]	[30, 10, 20]
indices	N/A	[1, 2, 0]	[1, 2, 0]	[1, 2, 0]
sorted_arr	N/A	N/A	[10, 20, 30]	[10, 20, 30]

Key Moments - 3 Insights

Why does np.argsort return indices instead of the sorted array?

np.argsort returns indices to show the order to rearrange the original array without changing it. This is shown in execution_table step 2, where indices [1, 2, 0] tell us how to reorder arr.

How do we get the sorted array from the indices?

By using the indices to index the original array, as in execution_table step 3, arr[indices] gives the sorted array [10, 20, 30].

What if the array has duplicate values?

np.argsort will return indices that sort duplicates in their original order (stable sort). This behavior is consistent with the indices shown in the example.

Visual Quiz - 3 Questions

Look at the execution_table at step 2, what does the indices array [1, 2, 0] represent?

A Positions to rearrange arr to sorted order

B Sorted values of arr

C Original array reversed

D Indices of maximum values

💡 Hint

Check the 'Indices Computed' column at step 2 in execution_table

At which step in the execution_table do we get the sorted array values?

A Step 2

B Step 3

C Step 1

D Step 4

💡 Hint

Look at the 'Result' column for the sorted array values

If the original array was [5, 5, 1], what would np.argsort(arr) return?

A [2, 1, 0]

B [0, 1, 2]

C [2, 0, 1]

D [1, 0, 2]

💡 Hint

Indices should sort the array: smallest value index first, duplicates keep original order

Concept Snapshot

np.argsort(array) returns indices that sort the array.
Use these indices to reorder the original array.
It does not sort the array itself.
Useful for sorting related arrays or stable sorting.
Example: indices = np.argsort(arr); sorted_arr = arr[indices]

Full Transcript

np.argsort() is a function in numpy that returns the indices which would sort an array. Instead of returning the sorted array directly, it gives you the order to rearrange the original array. For example, if you have an array [30, 10, 20], np.argsort returns [1, 2, 0] because the element at index 1 (10) is the smallest, then index 2 (20), then index 0 (30). You can then use these indices to get the sorted array by indexing the original array with them. This method is helpful when you want to sort multiple related arrays in the same order or keep track of the original positions. The execution table shows each step: starting with the array, computing indices, using them to reorder, and obtaining the sorted array. Key points include understanding that argsort returns indices, not sorted values, and how to use those indices to get sorted data.