AIF-C01 Question 191
Select 3A data science team is tasked with developing a machine learning model to predict customer churn. The team wants to ensure that the model is both transparent and explainable to stakeholders. They are considering using Amazon SageMaker tools and open-source resources for this purpose. Which of the following tools or practices would help them achieve this goal?
- A
Use Amazon SageMaker Model Cards to document the model's purpose, performance, and limitations.
- B
Use open-source interpretable models like linear regression or decision trees.
- C
Choose proprietary models with complex architectures to maximize accuracy.
- D
Review licensing and documentation for any open-source datasets and models used.
- E
Skip creating explainability reports to focus on improving model accuracy.
Show answer and explanation
Correct answers: A, B, D
Explanation
Transparent and explainable models are essential for building trust and ensuring ethical use of machine learning in business contexts. Amazon SageMaker Model Cards provide a structured way to document model information, while interpretable models and proper review of data and licensing enhance transparency. Combining these practices ensures stakeholders can understand and trust the model's outputs.
- A. Correct.
Amazon SageMaker Model Cards are specifically designed to document key details about the model, making it easier for stakeholders to understand its purpose, performance, and limitations. This supports transparency and explainability.
- B. Correct.
Interpretable models like linear regression or decision trees are inherently easier to understand and explain to stakeholders, making them a good choice for promoting model transparency.
- C. Incorrect.
While proprietary models with complex architectures can achieve high accuracy, they are often harder to explain to stakeholders due to their opaque nature. This option does not align with the goal of transparency and explainability.
- D. Correct.
Reviewing licensing and documentation ensures compliance and prevents issues with the use of open-source models and datasets. It also helps stakeholders understand the sources of the data and models, contributing to transparency.
- E. Incorrect.
Skipping explainability reports undermines the goal of transparency and makes it harder for stakeholders to trust the model. This is not a recommended practice.