Gradient Boosting (GBM) in ML Python - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is Gradient Boosting in simple terms?

Gradient Boosting is a way to build a strong prediction model by combining many weak models, usually decision trees, one after another. Each new model tries to fix the mistakes of the models before it.

beginner

How does Gradient Boosting improve model predictions step-by-step?

It builds models one at a time. Each model looks at the errors (residuals) from the previous models and learns to predict those errors better, reducing overall mistakes gradually.

intermediate

What role does the learning rate play in Gradient Boosting?

The learning rate controls how much each new model influences the final prediction. A smaller learning rate means the model learns slowly but can be more accurate, while a larger rate learns faster but might miss details.

beginner

Why are decision trees commonly used as weak learners in Gradient Boosting?

Decision trees are simple and fast to build. They can capture patterns in data easily and their errors can be corrected step-by-step, making them perfect for boosting.

intermediate

What is overfitting in Gradient Boosting and how can it be controlled?

Overfitting happens when the model learns the training data too well, including noise, and performs poorly on new data. It can be controlled by limiting tree depth, using a small learning rate, and stopping training early.

What does each new model in Gradient Boosting try to predict?

A The original target values directly

B Random noise in the data

C The errors made by previous models

D The average of all previous predictions

Each new model focuses on predicting the errors (residuals) from the combined previous models to improve accuracy.

Which of these is a common weak learner used in Gradient Boosting?

A Decision tree

B Neural network

C Support vector machine

D K-means clustering

Decision trees are simple models that work well as weak learners in Gradient Boosting.

What happens if the learning rate in Gradient Boosting is set too high?

A The model ignores previous errors

B The model will never learn anything

C The model will be very slow to train

D The model may learn too quickly and overfit

A high learning rate can cause the model to fit noise and overfit the training data.

How can overfitting be reduced in Gradient Boosting?

A By limiting tree depth and using a small learning rate

B By increasing the number of trees indefinitely

C By ignoring the residuals

D By using only one tree

Controlling tree complexity and learning rate helps prevent overfitting.

What is the main goal of Gradient Boosting?

A To reduce the size of the dataset

B To combine weak models to create a strong model

C To cluster data points into groups

D To randomly guess target values

Gradient Boosting builds many weak models that together make strong predictions.

Explain how Gradient Boosting builds a model step-by-step and why it focuses on errors.

Think about fixing mistakes one at a time.

Describe what overfitting means in Gradient Boosting and list two ways to prevent it.

Overfitting is like memorizing answers instead of understanding.