Elastic Net regularization in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding Elastic Net Regularization

Which statement best describes the Elastic Net regularization technique in machine learning?

A It randomly drops features during training to prevent overfitting.

B It uses only L1 penalty to shrink coefficients to zero, promoting feature selection.

C It applies only L2 penalty to reduce coefficient magnitude without setting any to zero.

D It combines L1 and L2 penalties to encourage both sparsity and grouping of correlated features.

💡 Hint

Think about how Elastic Net mixes two types of penalties.

Elastic Net regularization adds both L1 (lasso) and L2 (ridge) penalties to the loss function. This helps select important features (like L1) and also keeps correlated features grouped (like L2).

❓ Predict Output

intermediate

Output of Elastic Net Regression Coefficients

What will be the output coefficients after fitting Elastic Net regression with alpha=1.0 and l1_ratio=0.5 on the given data?

ML Python

from sklearn.linear_model import ElasticNet
import numpy as np

X = np.array([[1, 2], [2, 3], [3, 4], [4, 5]])
y = np.array([3, 5, 7, 9])

model = ElasticNet(alpha=1.0, l1_ratio=0.5, random_state=0)
model.fit(X, y)
coefficients = model.coef_
print(coefficients)

A [0.0 1.5]

B [0.5 1.0]

C [0.8 1.2]

D [1.0 0.0]

💡 Hint

Check how Elastic Net balances L1 and L2 penalties with given parameters.

With alpha=1.0 and l1_ratio=0.5, Elastic Net shrinks coefficients moderately. The output coefficients are approximately [0.5, 1.0].

❓ Hyperparameter

advanced

Effect of l1_ratio in Elastic Net

What happens to the Elastic Net model when the l1_ratio parameter is set to 1.0?

A The model applies both L1 and L2 penalties equally.

B The model behaves like Ridge regression, using only L2 penalty.

C The model behaves like Lasso regression, using only L1 penalty.

D The model ignores regularization and fits ordinary least squares.

💡 Hint

l1_ratio controls the mix between L1 and L2 penalties.

When l1_ratio=1.0, Elastic Net applies only the L1 penalty, making it equivalent to Lasso regression.

❓ Metrics

advanced

Choosing Regularization Strength Alpha

Which method is best to select the optimal alpha parameter for Elastic Net to balance bias and variance?

A Use cross-validation to find alpha that minimizes validation error.

B Set alpha to zero to avoid any regularization.

C Choose the largest alpha to maximize sparsity regardless of error.

D Randomly pick alpha without validation.

💡 Hint

Think about how to avoid overfitting and underfitting.

Cross-validation tests different alpha values and selects the one that gives the best performance on unseen data, balancing bias and variance.

🔧 Debug

expert

Debugging Elastic Net Model Training Error

You run this code but get a ValueError: 'alpha must be positive'. Which option fixes the error?

ML Python

from sklearn.linear_model import ElasticNet
model = ElasticNet(alpha=0, l1_ratio=0.5)
model.fit([[1, 2], [3, 4]], [5, 6])

A Change alpha to a positive value like 0.1.

B Set l1_ratio to 0 to disable L1 penalty.

C Remove the alpha parameter entirely.

D Change input data to integers only.

💡 Hint

Check the error message about alpha value.

ElasticNet requires alpha to be positive. Setting alpha=0 causes the error. Changing alpha to a positive number fixes it.