Objective 7.1 Report Findings CERTIFIED-DATA-SCIENCE-PRACTITIONER Practice Questions (Page 8)

Questions 36–40

1. 36foundation · easy

   In the context of reporting findings from a data science project, what is the primary purpose of user testing?

   • ATo evaluate whether the solution meets the needs and expectations of the end users
   • BTo measure the computational performance of the deployed model in production
   • CTo verify that the model's predictions are accurate on unseen data
   • DTo ensure the training data is free from bias and representative of the population
   Reveal Solution Discussion Select an answer first
2. 37application · medium

   A retail company's model predicts customer lifetime value. The data science team needs to identify which factors (e.g., purchase frequency, average order value, tenure) most strongly drive the predictions so the marketing team can focus their efforts. Which technique should the team use?

   • AGenerate a variable importance plot from the trained model.
   • BCompute the correlation matrix of all input features.
   • CPlot the distribution of the target variable.
   • DRun a principal component analysis and report the first component's loadings.
   Reveal Solution Discussion Select an answer first
4. 38expert · hard

   A team has trained a transformer-based model for document classification. A stakeholder asks why the model classified a particular document as 'urgent'. The team wants to show which words in the document most influenced the decision. Which explanation technique is most appropriate?

   • AUse LIME to generate a local surrogate model over the document's words.
   • BCompute SHAP values for the document's tokens.
   • CVisualize the attention weights from the model's attention layers, highlighting the words with the highest attention scores.
   • DReport the model's overall accuracy on the test set.
   Reveal Solution Discussion Select an answer first
5. 39application · medium

   A marketing team wants to use a propensity model to target the top 20% of customers most likely to respond to a campaign. The data scientist has produced a lift chart. Which interpretation of the lift chart correctly informs the targeting decision?

   • AA lift of 3.0 at the 20th percentile means the model has 3.0% higher accuracy than a random model on the top segment.
   • BIf the lift at the 20th percentile is 3.0, the model captures three times as many responders in that top segment as random selection would.
   • CThe lift chart should be read as the model's ROC curve, where a lift of 3.0 indicates an AUC of 0.80.
   • DIf the cumulative lift at 20% is 3.0, the model correctly predicts 30% of all responders within the top 20% of customers.
   Reveal Solution Discussion Select an answer first
6. 40foundation · easy

   Which of the following is an example of a post hoc explainable method?

   • ADecision tree
   • BLinear regression
   • CK-means clustering
   • DLIME
   Reveal Solution Discussion Select an answer first