Verify Data Quality for Go/no-Go Decision to Conduct Data Preparation PMI-CPMAI Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A healthcare analytics team has prepared a dataset of 50,000 patient records for a readmission-prediction model. The final data quality assessment shows 99.8% completeness for all features, no duplicate records, and consistent date formats. However, the dataset contains only patients from urban hospitals, while the model will be deployed nationally. The team lead must decide whether to proceed to model training. What should the team lead do?

   • ADelay the go decision and collect or acquire rural hospital data before training.
   • BProceed to training because completeness and consistency exceed the 99% threshold.
   • CProceed to training and plan to fine-tune the model later with rural data.
   • DProceed to training but add a note in the model card about the urban-only limitation.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   After applying a standard scaling transformation to numeric features, what is the expected output for each transformed feature?

   • AMean of 0 and standard deviation of 1
   • BMedian of 0 and interquartile range of 1
   • CAll values become integers
   • DValues between 0 and 1
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A healthcare company is building a model to predict patient no-shows. The training data is from a single clinic that serves a predominantly urban population. The model will be deployed across the company's network of clinics, including rural locations. The data quality assessment shows high completeness and consistency. What is the most important factor to consider before the go/no-go decision?

   • AThe data is timely because it is from the current year, so the model is ready.
   • BThe data is complete and consistent, so the model is ready for training.
   • CThe data is not representative of the rural clinics, so the model may not generalize well.
   • DThe data is accurate because it comes from a single clinic, so the model is ready.
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A government agency is developing a model to predict unemployment benefit fraud. The training data is sourced from historical claims, but the agency knows that fraud detection was less effective in previous years, meaning some fraudulent claims were never flagged. The model will be used to flag new claims for review. What is the primary bias concern and the most appropriate mitigation?

   • AMeasurement bias because the fraud detection process was inconsistent; standardize the detection process first.
   • BNo bias concern because the model will learn the patterns from the historical data.
   • CLabel bias because the historical labels are unreliable; consider using active learning or manual review to create a more accurate label set.
   • DSampling bias because the data is too old; collect more recent claims data before training.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   Which factor is NOT typically considered when making a go/no-go decision on data readiness?

   • ARepresentativeness of the dataset
   • BPotential bias in the data
   • CData quality dimensions such as completeness and accuracy
   • DThe programming language used for model development
   Reveal Solution Discussion Select an answer first