How to Use np.argsort in NumPy: Syntax and Examples

Use np.argsort to get the indices that would sort an array. It returns an array of indices that sort the original array in ascending order by default. You can use these indices to reorder the array or related data.

Syntax

The basic syntax of np.argsort is:

• np.argsort(a, axis=-1, kind='quicksort', order=None)

Where:

• a: The input array to sort.
• axis: The axis along which to sort. Default is the last axis.
• kind: Sorting algorithm to use (e.g., 'quicksort', 'mergesort', 'heapsort', 'stable').
• order: When sorting structured arrays, specifies field names to sort by.

python

np.argsort(a, axis=-1, kind='quicksort', order=None)

Example

This example shows how to use np.argsort to get the indices that sort an array, and then use those indices to sort the array.

python

import numpy as np

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

print('Original array:', arr)
print('Indices that sort the array:', indices)
print('Sorted array using indices:', sorted_arr)

Output

Original array: [30 10 20]
Indices that sort the array: [1 2 0]
Sorted array using indices: [10 20 30]

Common Pitfalls

One common mistake is expecting np.argsort to return the sorted array itself. Instead, it returns the indices that would sort the array. You must use these indices to reorder the array.

Another pitfall is not specifying the correct axis when working with multi-dimensional arrays, which can lead to unexpected results.

python

import numpy as np

arr = np.array([30, 10, 20])

# Wrong: expecting argsort to return sorted array
sorted_wrong = np.argsort(arr)
print('Wrong output:', sorted_wrong)  # This is indices, not sorted values

# Right: use indices to sort
sorted_right = arr[np.argsort(arr)]
print('Correct sorted array:', sorted_right)

Output

Wrong output: [1 2 0]
Correct sorted array: [10 20 30]

Quick Reference

Parameter	Description	Default
a	Input array to sort	Required
axis	Axis along which to sort	-1 (last axis)
kind	Sorting algorithm ('quicksort', 'mergesort', 'heapsort', 'stable')	'quicksort'
order	Field names for structured arrays	None

Key Takeaways

np.argsort returns indices that would sort an array, not the sorted array itself.

Use the returned indices to reorder the original array or related data.

Specify the axis parameter carefully when sorting multi-dimensional arrays.

Common sorting algorithms include 'quicksort', 'mergesort', and 'heapsort'.

For structured arrays, use the order parameter to sort by specific fields.