Model Iteration DY0-001 Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A data science team is building a binary classification model to detect spam emails. The dataset has 10,000 emails, with 20% being spam. The team trained a model and is evaluating it on a test set. The confusion matrix shows: True Positives (spam correctly identified) = 150, False Negatives (spam missed) = 50, True Negatives (non-spam correctly identified) = 7,800, False Positives (non-spam flagged as spam) = 2,000. The business wants to minimize the number of legitimate emails that are incorrectly flagged as spam. Which metric should the team focus on?

   • AF1-score, because it balances precision and recall
   • BAccuracy, because it measures the overall correctness of the model
   • CPrecision, because it measures the proportion of predicted spam emails that are actually spam
   • DRecall, because it measures the proportion of spam emails correctly identified
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A credit risk team is developing a model to predict loan default. They are using a dataset with 100,000 records and 50 features. The team has split the data into training and test sets. They are considering using k-fold cross-validation to evaluate the model. What is the primary benefit of using k-fold cross-validation over a single train-test split?

   • AIt guarantees the model will not overfit the training data
   • BIt reduces the need for feature engineering
   • CIt eliminates the need for a separate test set
   • DIt provides a more robust estimate of model performance by using all data for both training and validation across multiple folds
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   A data scientist is designing a model to predict customer churn. The business requires the model to be interpretable so that marketing teams can understand why customers are predicted to leave. Which model design principle is most directly prioritized in this scenario?

   • AEnsuring the model uses all available features
   • BMinimizing training time above all other considerations
   • CMaximizing predictive accuracy regardless of complexity
   • DBalancing model complexity with interpretability
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A data science team is developing a model to predict loan default. The dataset has 200,000 records and 100 features. The team has split the data into training (70%) and test (30%) sets. They are using logistic regression and have achieved an AUC of 0.78 on the test set. The business wants to improve the model's performance. The team is considering two approaches: (1) adding polynomial features to capture non-linear relationships, or (2) using a random forest model. The business has a strict requirement that the model must be interpretable for regulatory compliance. Which approach should the team take?

   • AAdd polynomial features, since they always improve model performance
   • BUse a random forest model, since it can be explained using SHAP values
   • CAdd polynomial features to the logistic regression, since it maintains interpretability while capturing non-linear relationships
   • DUse a random forest model, since it will likely achieve higher AUC than logistic regression
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   A data scientist has trained a binary classification model and wants to evaluate its performance on a test set. Which metric is most appropriate when the cost of false negatives is much higher than the cost of false positives?

   • AF1 score
   • BPrecision
   • CAccuracy
   • DRecall (sensitivity)
   Reveal Solution Discussion Select an answer first