Gradient Boosting (GBM) in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

How does Gradient Boosting improve model predictions?

Gradient Boosting builds a strong model by combining many weak models. What is the main way it improves predictions at each step?

A It fits a new model to the residual errors of the previous model to reduce overall error.

B It randomly selects features to build each new model independently.

C It averages the predictions of all previous models without adjustment.

D It increases the depth of the decision trees in each iteration.

💡 Hint

Think about how the model learns from mistakes made before.

Gradient Boosting fits each new model to the errors (residuals) of the combined previous models, improving predictions step-by-step.

❓ Predict Output

intermediate

Output of training loss during Gradient Boosting

Consider this Python snippet training a Gradient Boosting Regressor on a simple dataset. What is the printed training loss after 3 iterations?

ML Python

from sklearn.ensemble import GradientBoostingRegressor
from sklearn.metrics import mean_squared_error
import numpy as np

X = np.array([[1], [2], [3], [4], [5]])
y = np.array([1.5, 3.5, 3.0, 5.0, 7.5])

model = GradientBoostingRegressor(n_estimators=3, learning_rate=1.0, max_depth=1, random_state=42)
model.fit(X, y)
pred = model.predict(X)
loss = mean_squared_error(y, pred)
print(round(loss, 2))

A 0.50

B 0.00

C 0.12

D 1.25

💡 Hint

Check how well the model fits the small dataset after 3 boosting steps.

The model reduces error progressively; after 3 estimators, the mean squared error is about 0.12.

❓ Model Choice

advanced

Choosing the best base learner for Gradient Boosting

You want to use Gradient Boosting for a classification task with many categorical features. Which base learner is most suitable?

A Linear regression models

B Decision trees with limited depth

C K-nearest neighbors

D Support vector machines

💡 Hint

Think about what base learners are commonly used in Gradient Boosting and handle categorical data well.

Decision trees are the standard base learners in Gradient Boosting and handle categorical features effectively.

❓ Hyperparameter

advanced

Effect of learning rate in Gradient Boosting

What happens if you set the learning rate too high in a Gradient Boosting model?

A The model ignores residuals and stops learning.

B The model automatically prunes trees to prevent overfitting.

C The model will train slower but generalize better.

D The model may overfit quickly and have unstable training.

💡 Hint

Consider how a large step size affects the model updates.

A high learning rate causes large updates that can lead to overfitting and unstable training.

❓ Metrics

expert

Interpreting feature importance in Gradient Boosting

After training a Gradient Boosting model, you get these feature importances: {'age': 0.6, 'income': 0.3, 'gender': 0.1}. What does this mean?

A The 'age' feature contributes most to reducing prediction error in the model.

B All features contribute equally to the model's predictions.

C The 'gender' feature is the most important for predictions.

D Feature importance values indicate the correlation between features.

💡 Hint

Think about what feature importance measures in Gradient Boosting.

Feature importance shows how much each feature helps reduce error; higher means more contribution.