ML Workflow CT-AI Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A data scientist has trained a regression model and wants to evaluate its performance. The dataset has some outliers that may disproportionately affect the error metric. Which metric is most robust to outliers?

   • ARoot mean squared error (RMSE)
   • BMean squared error (MSE)
   • CR-squared
   • DMean absolute error (MAE)
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A company has deployed a machine learning model to predict equipment failure in real-time. After a few months, the model's performance degrades because the equipment behavior changes over time. What is the most appropriate action to maintain model performance?

   • AFreeze the model to ensure consistent predictions
   • BReduce the frequency of predictions to lower the impact of changes
   • CIncrease the model's complexity to capture more patterns
   • DImplement continuous monitoring and retrain the model periodically with new data
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   What is the purpose of using a test set in model evaluation?

   • ATo clean the data by removing outliers
   • BTo train the model on more data
   • CTo tune the model's hyperparameters
   • DTo provide an unbiased assessment of the final model's performance
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which of the following best describes the typical sequence of steps in an ML workflow?

   • AData collection → data preparation → model training → model evaluation → model deployment and monitoring
   • BModel evaluation → data collection → model training → data preparation → model deployment
   • CData preparation → model deployment → model training → data collection → model evaluation
   • DModel training → data collection → model evaluation → data preparation → model deployment
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A data scientist is building a model to predict customer churn. The dataset has a feature 'customer tenure' with many missing values. The scientist also plans to use a model that requires all features to be numeric. The missing values are not random; customers with longer tenure are more likely to have missing values. Which approach is most appropriate?

   • AImpute missing values using the median tenure, and add a binary indicator for missingness
   • BImpute missing values with the mean tenure without an indicator
   • CReplace missing values with zero
   • DRemove all rows with missing tenure values
   Reveal Solution Discussion Select an answer first