MLS-C01 Question 123
Select 3You are building a machine learning model to predict customer churn for a subscription-based service. After splitting the dataset into training and test sets, you observe that the model performs well on the training data but poorly on the test data. Which of the following steps can help address this issue?
- A
Increase the complexity of the model by adding more layers or parameters.
- B
Apply regularization techniques such as L1 or L2 regularization.
- C
Collect more data to increase the size of the training dataset.
- D
Use cross-validation to better estimate the model's generalization performance.
- E
Reduce the number of features in the dataset to remove irrelevant or redundant features.
Show answer and explanation
Correct answers: B, D, E
Explanation
The scenario describes a classic case of overfitting, where the model performs well on the training data but poorly on the test data. To address this, techniques such as regularization (L1/L2), cross-validation, and feature selection can be used to reduce overfitting and improve generalization. Collecting more data might help but is not an immediate solution, and increasing model complexity would likely worsen the problem.
- A. Incorrect.
Increasing the complexity of the model might exacerbate overfitting, as the model is already performing poorly on the test data despite doing well on the training data. This is not a suitable step to address the issue.
- B. Correct.
Applying regularization techniques such as L1 or L2 can penalize overly complex models, reducing overfitting and improving test set performance.
- C. Incorrect.
Collecting more data can help improve model generalization, but it is not always feasible or practical in the short term. In this scenario, other approaches should be prioritized first.
- D. Correct.
Using cross-validation provides a better understanding of the model's generalization performance and helps in selecting hyperparameters that reduce overfitting.
- E. Correct.
Reducing the number of features can eliminate irrelevant or redundant information, which may help the model generalize better and reduce overfitting.