Task 2.3: Analyze Model Performance MLA-C01 Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A bank is using SageMaker Clarify to analyze a binary classification model that predicts loan approval. The model uses features including age, income, credit score, and zip code. The data scientist runs a Clarify bias analysis and receives a report showing that the model's predicted approval rate for applicants from a certain zip code is significantly lower than for other zip codes. What is the most appropriate next step?

   • AIncrease the model's complexity to better capture the relationship between zip code and loan approval.
   • BImmediately remove the zip code feature from the model to eliminate bias.
   • CIgnore the finding because zip code is a legitimate business factor for loan approval.
   • DInvestigate the training data for label bias or sampling bias in that zip code before deciding on mitigation.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   Which fairness metric measures the difference in the false positive rate between two groups?

   • AEqualized odds
   • BClass imbalance
   • CDemographic parity
   • DLabel skew
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   What is a typical use case for Amazon SageMaker Debugger's built-in rules?

   • ACreating confusion matrices
   • BDetecting when the loss is not decreasing
   • CGenerating SHAP values for predictions
   • DMonitoring data drift in production
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   Which SageMaker Clarify feature is used to detect bias in a trained model's predictions?

   • AData Quality Report
   • BModel Monitor
   • CModel Explainability
   • DBias Report
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   During training, the loss function plateaus at a high value and does not decrease further. What is a common technique to address this convergence issue?

   • AAdjust the learning rate schedule or use a different optimizer
   • BDecrease the batch size
   • CIncrease the learning rate
   • DIncrease the number of epochs
   Reveal Solution Discussion Select an answer first