A data scientist is training a regression model to predict house prices in Azure Machine Learning. The model uses features like square footage, number of bedrooms, and location (zip code). The data scientist notices that the model has a very low error on the training data but a high error on the test data. Which technique should the data scientist apply during model training to reduce overfitting by penalizing large coefficients?

Use a smaller test set.

A smaller test set would provide a less reliable evaluation of generalization and does not address overfitting during training.

Apply feature scaling only.

Feature scaling (e.g., normalization) helps gradient descent converge but does not penalize large coefficients or directly reduce overfitting.

Increase the number of training epochs.

More training epochs can cause the model to overfit further by learning the training data more precisely, increasing the gap between training and test error.

What does this AI-900 question test?

Read the scenario before looking for a memorised answer.

What is the correct answer to this question?

The correct answer is: Use a regularization algorithm like Lasso (L1). — Overfitting occurs when the model learns noise in the training data, often indicated by low training error and high test error. Regularization techniques like Lasso (L1) or Ridge (L2) add a penalty term for large coefficients, which discourages complexity and reduces overfitting. Using a smaller test set (A) would make evaluation less reliable. Feature scaling (B) helps optimization but does not directly penalize coefficients. Increasing training epochs (D) can actually worsen overfitting by allowing the model to fit the training data more precisely.

What should I do if I get this AI-900 question wrong?

Then try more questions from the same exam bank and focus on understanding why the wrong options are tempting.