Describe Best Practices for Communicating Findings and Operationalizing an Analytics Project DATA-SCIENCE-FOUNDATIONS Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A data science team has built a model to predict customer lifetime value (CLV). They need to communicate the results to the sales team, who will use the model to prioritize leads. What is the most effective way to present the findings?

   • APresent a detailed technical report explaining the model's algorithm, feature importance, and validation metrics.
   • BGive the sales team the raw data file with all customer records and predicted CLV values, and let them sort it themselves.
   • CShow a series of complex charts that demonstrate the model's predictive power, such as ROC curves and lift charts.
   • DProvide a simple dashboard that ranks customers by predicted CLV, with clear labels and a brief explanation of how to use the ranking.
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A company has developed a fraud detection model in a Jupyter notebook and wants to deploy it to production so that transaction data can be scored in real time. Which step is the most critical first consideration for operationalizing this model?

   • ARefactor the model code into a production-grade service with a defined API, input validation, and logging.
   • BCreate a detailed PowerPoint presentation for stakeholders explaining the model's architecture and expected performance.
   • CRetrain the model on all historical data one final time to ensure it is as accurate as possible before deployment.
   • DImmediately connect the notebook to the production transaction database to start scoring live traffic.
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   A data science team has developed a model that predicts employee attrition. The HR director wants to use the model to identify at-risk employees and intervene. The data science team is concerned that the model's predictions could be seen as biased or intrusive. What is the best way to communicate the model's findings to the HR director and address these concerns?

   • ARecommend that the HR director not use the model at all, citing the potential for bias and the intrusive nature of the predictions.
   • BPresent only the model's overall accuracy and a list of at-risk employees, without discussing the underlying factors or limitations.
   • CPresent the model's findings with a clear explanation of the factors driving predictions, and discuss the ethical implications and limitations of using the model for HR decisions.
   • DFocus the presentation on the technical sophistication of the model to build confidence in its predictions.
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A retail company is deploying a model to recommend products in real time on its website. The model must respond within 100 milliseconds. The data science team has built a complex ensemble model that is highly accurate but takes 500 milliseconds to generate a prediction. What is the best approach to meet the latency requirement while maintaining as much accuracy as possible?

   • ADeploy the complex model as-is and accept the 500-millisecond latency, since accuracy is the most important factor.
   • BSimplify the model to a faster, less complex algorithm, and evaluate the trade-off between latency and accuracy to find an acceptable balance.
   • CIncrease the computing resources allocated to the model to speed up its execution time.
   • DUse a caching mechanism to store predictions for common requests, so the model does not need to run for every request.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   What is the primary purpose of using narrative techniques in data storytelling?

   • ATo guide the audience through the insights in a logical and engaging way
   • BTo replace the need for any visualizations
   • CTo make the data more complex and impressive
   • DTo hide any limitations of the analysis
   Reveal Solution Discussion Select an answer first