MLA-C01 Question 175
Select 3You are training a machine learning model to predict user churn for a subscription service using Amazon SageMaker. During evaluation, the model shows low accuracy and high bias on the training and validation datasets. Which steps should you take to improve the model's performance?
- A
Increase the size of the training dataset by incorporating more labeled data.
- B
Use a more complex model architecture that can better capture data patterns.
- C
Apply feature engineering techniques, such as scaling or encoding categorical variables.
- D
Reduce the learning rate of your optimizer to improve convergence during training.
- E
Add dropout layers to the model to prevent overfitting.
Show answer and explanation
Correct answers: A, B, C
Explanation
To address low accuracy and high bias, you should focus on increasing the diversity of the training data, improving the model architecture, and optimizing feature engineering. These steps help the model generalize better and learn the underlying patterns in the data. Reducing the learning rate and adding dropout layers are more relevant for scenarios involving overfitting or optimization convergence.
- A. Correct.
Increasing the training dataset size can help the model generalize better and reduce bias by providing more diverse examples.
- B. Correct.
Using a more complex model architecture can help capture intricate patterns in the data, which may reduce bias and improve performance.
- C. Correct.
Feature engineering, such as scaling or encoding categorical variables, ensures that the input data is properly prepared for the model, improving its performance.
- D. Incorrect.
Reducing the learning rate helps fine-tune convergence, but it is not a direct solution for addressing low accuracy or high bias.
- E. Incorrect.
Adding dropout layers is a technique to reduce overfitting, but the scenario describes high bias rather than high variance, so dropout would not address the root cause.