Answer-first summary for fast verification
Answer: Create alerts to monitor for skew, and retrain the model.
The correct answer is A. When the distribution of input data changes, it can lead to a drop in the model's performance. This situation is known as data drift or skew. To address this, it is important to monitor the input data for skew. By creating alerts, you can detect when the distribution of the input data changes. Once detected, retraining the model with the new data that reflects the updated distribution can help maintain or restore the model's performance. Regularization (as suggested in option C) and feature selection (as suggested in options B and D) address specific aspects of the model but do not directly monitor changes in data distribution.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
No comments yet.
Your team recently deployed a Deep Neural Network (DNN) regression model, which had shown good performance during training and testing phases. However, six months post-deployment, you observe a significant drop in the model's performance. This decline is attributed to a change in the distribution of the input data, indicating data drift. To address the differences in input data distribution and restore the model's performance in production, what steps should you take?
A
Create alerts to monitor for skew, and retrain the model.
B
Perform feature selection on the model, and retrain the model with fewer features.
C
Retrain the model, and select an L2 regularization parameter with a hyperparameter tuning service.
D
Perform feature selection on the model, and retrain the model on a monthly basis with fewer features.