Why broadcasting matters in NumPy - Challenge Your Understanding

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of adding arrays with broadcasting

What is the output of this code snippet using NumPy broadcasting?

NumPy

import numpy as np

arr1 = np.array([1, 2, 3])
arr2 = np.array([[10], [20], [30]])
result = arr1 + arr2
print(result)

A ValueError: operands could not be broadcast together with shapes (3,) (3,1)

B

[[11 12 13]
 [21 22 23]
 [31 32 33]]

C

[[10 20 30]
 [11 21 31]
 [12 22 32]]

D

[[11 12 13 10]
 [21 22 23 20]
 [31 32 33 30]]

💡 Hint

Think about how NumPy aligns shapes when adding arrays of different dimensions.

NumPy broadcasts the 1D array arr1 across each row of the 2D arr2, adding element-wise to produce a 3x3 array.

❓ data_output

intermediate

Shape of result after broadcasting

Given these arrays, what is the shape of the result after broadcasting and addition?

NumPy

import numpy as np

x = np.ones((4,1))
y = np.arange(3)
z = x + y

A (4, 3)

B (4, 1)

C (3, 4)

D (1, 3)

💡 Hint

Broadcasting expands dimensions to match the largest shape.

x has shape (4,1) and y has shape (3,). y is treated as (1,3). Broadcasting results in (4,3).

🔧 Debug

advanced

Identify the broadcasting error

Why does this code raise an error?

NumPy

import numpy as np

arr1 = np.array([1, 2])
arr2 = np.array([[1, 2, 3], [4, 5, 6]])
result = arr1 + arr2

A IndexError: index out of bounds

B TypeError: unsupported operand type(s) for +: 'int' and 'list'

C ValueError: operands could not be broadcast together with shapes (2,) (2,3)

D No error, result is [[2 4 4] [5 7 8]]

💡 Hint

Check if the shapes of the arrays can be broadcast together.

arr1 shape is (2,) and arr2 shape is (2,3). These shapes are incompatible for broadcasting.

🚀 Application

advanced

Using broadcasting to normalize data

You have a 2D array of data where each column needs to be normalized by subtracting the column mean. Which code correctly uses broadcasting to do this?

NumPy

import numpy as np
data = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
# Normalize columns by subtracting mean

A data - data.mean(axis=0, keepdims=True)

B data - data.mean(axis=1)

C data - data.mean()

D data - data.mean(axis=0)

💡 Hint

Consider the shape of the mean array and how broadcasting works for subtraction.

Using keepdims=True keeps the mean shape compatible for broadcasting across rows.

🧠 Conceptual

expert

Why broadcasting improves performance

Why does NumPy use broadcasting instead of explicit loops for operations on arrays of different shapes?

A Broadcasting forces arrays to be the same shape by copying data multiple times.

B Broadcasting allows Python to automatically parallelize code across multiple CPUs.

C Broadcasting converts arrays to lists internally for faster iteration.

D Broadcasting avoids creating large temporary arrays, saving memory and speeding up computation.

💡 Hint

Think about memory usage and speed when working with large data.

Broadcasting lets NumPy perform operations without copying data unnecessarily, improving speed and memory use.