Polynomial features in ML Python - Cheat Sheet & Quick Revision

Recall & Review

beginner

What are polynomial features in machine learning?

Polynomial features are new features created by raising existing features to a power and combining them. They help models learn more complex patterns by adding curved relationships.

beginner

Why do we use polynomial features in a model?

We use polynomial features to allow simple models like linear regression to fit curves and capture non-linear relationships in data.

intermediate

How does degree affect polynomial features?

The degree controls the highest power of features included. Higher degree means more complex features but can cause overfitting if too high.

beginner

What is an example of polynomial features for two variables x and y with degree 2?

For x and y, degree 2 polynomial features include: 1 (bias), x, y, x², xy, y².

intermediate

What is a potential downside of using polynomial features?

Using too many polynomial features can make the model complex, slow to train, and prone to overfitting, especially with small data.

What does adding polynomial features to a dataset help a model do?

A Reduce the number of features

B Make the model linear only

C Capture non-linear relationships

D Remove noise from data

Polynomial features help models learn curved or non-linear patterns by adding powers and combinations of original features.

Which of these is NOT a polynomial feature of degree 2 for variables x and y?

A y³

B xy

C x²

D y²

y³ is degree 3, which is higher than degree 2, so it is not included in degree 2 polynomial features.

What happens if you use a very high degree for polynomial features?

A Model becomes simpler

B Model may overfit

C Model ignores features

D Model always improves

High degree polynomial features can cause the model to fit noise, leading to overfitting.

Polynomial features are mainly used with which type of model?

A Linear models

B Decision trees

C Neural networks only

D Clustering algorithms

Polynomial features transform data so linear models can fit non-linear patterns.

Which term is usually included automatically when generating polynomial features?

A Only squared features

B Only original features

C Only interaction terms

D Bias term (constant 1)

The bias term (constant 1) is included to allow the model to fit an intercept.

Explain what polynomial features are and why they are useful in simple terms.

Think about how adding powers of features changes what the model can learn.

Describe the risks of using very high degree polynomial features in a model.

Consider what happens when a model tries to fit too many details.