Fancy Indexing in NumPy

Fancy indexing enables advanced and flexible way to retrieve data from the numpy array. 

By using fancy indexing we use a very simple syntax to get any data from the array.

Syntax:

array[[row_indices], [column_indices]]

Example or problem (Let's say we have an numpy array:) :

import numpy as np

x = np.array([[12, 34, 45, 45], [34, 5, 1, 63], [12, 23, 6, 34], [111, 222, 2, 333]])

print(x)

[[ 12  34  45  45]
 [ 34   5   1  63]
 [ 12  23   6  34]
 [111 222   2 333]]

Question is : Retrieve the first and fourth rows and second and third columns only using indexing:

Solution:

 Now, with the help of simple indexing, it looks like impossible to retrieve this data. So let's do this with fancy indexing:

import numpy as np

x = np.array([[12, 34, 45, 45], [34, 5, 1, 63], [12, 23, 6, 34], [111, 222, 2, 333]])

row_indexes = [0, 2]

print(x[row_indexes][:])

Note : Note: Remember, when we use : this symbol, it means all rows or columns.

Output: By this we simply get 0 index and 2 nd index rows.     

[[12 34 45 45]
 [12 23  6 34]]

Not we target the columns only to complete the problem: 

import numpy as np

x = np.array([[12, 34, 45, 45], [34, 5, 1, 63], [12, 23, 6, 34], [111, 222, 2, 333]])

column_indexes = [1, 2]

print(x[:, column_indexes])

 Output: By this we simply get 1 index and 2 nd index columns.     

[[ 34  45]
 [  5   1]
 [ 23   6]
 [222   2]]

Let's merge these two queries to achieve the desired result.

import numpy as np

x = np.array([[12, 34, 45, 45], [34, 5, 1, 63], [12, 23, 6, 34], [111, 222, 2, 333]])

row_indexes = [0, 2]
column_indexes = [1, 2]

print(x[row_indexes, :][:, column_indexes])

 Output: By merging these two queries we get our result :    

[[34 45]
 [23  6]]

Let's do one more example:

We have a numpy array:

import numpy as np

pq = np.array([[6, 113,  83, 115, 105], [110,  83,  76,  96, 18], [ 33,  76,   5,  84,  58], [116, 105,  11,  84,  69]])

print(pq)

From this numpy array, i need to get this data:

[[  6 113  83 115 105]
 [110  83  76  96 18]
 [ 33  76   5  84  58]
 [116 105  11  84  69]]

Solution:

import numpy as np

pq = np.array([[6, 113,  83, 115, 105], [110,  83,  76,  96, 18], [ 33,  76,   5,  84,  58], [116, 105,  11,  84,  69]])

print(pq[[1, 2, 3], :][:, [1, 2, 3]])

Output:

[[ 83  76  96]
 [ 76   5  84]
 [105  11  84]]

15+ Fancy Indexing Interview Questions in Numpy

Question 1: What is fancy indexing in NumPy?

Answer: Fancy indexing is a technique in NumPy that allows accessing array elements using an array or list of integer indices, rather than slicing with simple ranges.

Example:

import numpy as np

arr = np.array([10, 20, 30, 40, 50])
indices = [0, 2, 4]

print(arr[indices])

Output:

[10 30 50]

Question 2: How is fancy indexing different from regular slicing?

Answer: Regular slicing uses a range or step (:) to extract data, while fancy indexing uses specific indices in the form of arrays or lists.

Example:

import numpy as np

arr = np.array([10, 20, 30, 40, 50])
print(arr[1:4])  # Slices a range

# Fancy Indexing
indices = [0, 2, 4]
print(arr[indices])  # Selects specific elements

Output:

[20 30 40]
[10 30 50]

Question 3: Can fancy indexing be used on multidimensional arrays?

Answer: Yes, fancy indexing works with multidimensional arrays.

Example:

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
row_indices = [0, 2]
col_indices = [1, 2]
print(arr[row_indices, col_indices])

Output:

[2 9]

Question 4: How can you retrieve rows and columns using fancy indexing?

Answer: To retrieve specific rows and columns, you combine fancy indexing for rows and columns.

Example:

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
rows = [0, 2]
cols = [1, 2]
print(arr[rows][:, cols])

Output:

[[2 3]
 [8 9]]

Question 5: What happens if you use out-of-bound indices in fancy indexing?

Answer: Using out-of-bound indices raises an IndexError.

Example:

import numpy as np

arr = np.array([1, 2, 3, 4])
indices = [0, 5]  # 5 is out-of-bounds
print(arr[indices])  # Uncommenting will raise an error

Output:

IndexError: index 5 is out of bounds for axis 0 with size 4

Question 6: Can you mix slicing and fancy indexing?

Answer: Yes, slicing and fancy indexing can be mixed in the same operation.

Example:

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
print(arr[1:, [0, 2]])

Output:

[[4 6]
 [7 9]]

Question 7: How do you use fancy indexing to access all rows but specific columns?

Answer: Use a slice for rows (:) and fancy indexing for columns.

Example:

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
print(arr[:, [0, 2]])

Output:

[[1 3]
 [4 6]
 [7 9]]

Question 8: Can you assign values using fancy indexing?

Answer: Yes, you can assign values to specific elements using fancy indexing.

Example:

[ 10 200  30 400  50]

Output:

[[ 83  76  96]
 [ 76   5  84]
 [105  11  84]]

Question 9: How can fancy indexing be used to retrieve diagonal elements?

Answer: Diagonal elements can be accessed by providing matching row and column indices.

Example:

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
print(arr[[0, 1, 2], [0, 1, 2]])

Output:

[1 5 9]

Question 10: Is it possible to use boolean arrays with fancy indexing?

Answer: Yes, boolean arrays can also be used to filter elements.

Example:

import numpy as np

arr = np.array([10, 20, 30, 40, 50])
mask = [True, False, True, False, True]
print(arr[mask])

Output:

[10 30 50]

Question 11: Can you use negative indices in fancy indexing?

Answer: Yes, negative indices work in fancy indexing, referring to elements from the end.

Example:

import numpy as np

arr = np.array([10, 20, 30, 40, 50])
print(arr[[-1, -3]])

Output:

[50 30]

Question 12: How do you combine rows and columns with fancy indexing?

Answer: Provide row and column indices together in a tuple.

Example:

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
rows = [0, 1]
cols = [2, 0]
print(arr[rows, cols])

Output:

[3 4]

Question 13: How can you use fancy indexing to rearrange array elements?

Answer: Fancy indexing can rearrange array elements by specifying indices in the desired order.

Example:

import numpy as np

arr = np.array([10, 20, 30, 40, 50])
new_order = [4, 0, 3, 1, 2]
print(arr[new_order])

Output:

[50 10 40 20 30]

Question 14: What is the result of x[row_indices][:, column_indices] in fancy indexing?

Answer: It retrieves elements by first slicing rows, then applying fancy indexing to columns.

Example:

import numpy as np

x = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
rows = [0, 2]
cols = [1, 2]
print(x[rows][:, cols])

Output:

[[2 3]
 [8 9]]

Question 15: Can fancy indexing be applied to 3D arrays?

Answer: Yes, fancy indexing works for 3D arrays.

Example:

import numpy as np

arr = np.array([[[1, 2], [3, 4]], [[5, 6], [7, 8]]])
print(arr[[0, 1], [0, 1], [1, 0]])

Output:

[2 7]