Avoiding broadcasting mistakes in NumPy - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is broadcasting in numpy?

Broadcasting is numpy's way to perform operations on arrays of different shapes by automatically expanding the smaller array to match the larger one without copying data.

beginner

Why can broadcasting cause mistakes in numpy calculations?

Broadcasting can cause mistakes when arrays have shapes that seem compatible but lead to unintended expansion, causing wrong results or hard-to-find bugs.

beginner

How can you check the shapes of arrays before broadcasting?

You can print the shapes using the `.shape` attribute of numpy arrays to ensure they align as expected before performing operations.

intermediate

What is a safe way to avoid broadcasting mistakes when adding arrays?

Make sure arrays have compatible shapes or explicitly reshape arrays using `.reshape()` or `np.expand_dims()` to control how broadcasting happens.

beginner

What happens if you try to broadcast arrays with incompatible shapes?

Numpy raises a ValueError indicating the shapes are not compatible for broadcasting, preventing silent wrong calculations.

What does numpy do when you add a (3,1) array to a (3,4) array?

A Add only the first column of the (3,4) array

B Raise an error because shapes differ

C Broadcast the (3,4) array to (3,1)

D Broadcast the (3,1) array to (3,4) and add element-wise

Numpy broadcasts the smaller dimension (1) to match 4, so (3,1) becomes (3,4) for element-wise addition.

Which shape pair will cause a broadcasting error?

A (3,2) and (2,3)

B (4,1) and (4,5)

C (2,3) and (3,)

D (5,) and (5,)

Shapes (3,2) and (2,3) are incompatible for broadcasting because their dimensions do not align from the right.

How can you avoid unexpected broadcasting results?

A Always check array shapes before operations

B Use Python lists instead of numpy arrays

C Ignore shapes and trust numpy

D Only use arrays of the same shape

Checking shapes helps ensure arrays will broadcast as intended, avoiding mistakes.

What numpy method helps to add a new axis for broadcasting?

A np.concatenate()

B np.flatten()

C np.expand_dims()

D np.transpose()

np.expand_dims() adds a new axis, helping control broadcasting behavior.

If you want to add a (3,) array to each row of a (3,4) array, what should you do?

A Transpose the (3,4) array first

B Reshape (3,) to (3,1) before adding

C No need to reshape, just add directly

D Reshape (3,) to (1,3) before adding

Reshaping (3,) to (3,1) aligns dimensions for broadcasting across columns.

Explain what broadcasting is and why it can cause mistakes in numpy.

Think about how numpy handles arrays of different shapes during operations.

Describe steps you can take to avoid broadcasting mistakes when working with numpy arrays.

Consider how to prepare arrays before operations.