Determine If Data Meets Solution Needs PMI-CPMAI Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   A manufacturing company is building a quality control model to detect defects. The data science team has images of products, but only 2% are defective. The model will be used to automatically reject defective products. The team is assessing data sufficiency. What should they do?

   • AProceed with the data because the total number of images is large.
   • BAssess whether the number of defective images is sufficient to train a model that can reliably detect defects.
   • CProceed with the data and use a threshold adjustment to improve defect detection.
   • DCollect more non-defective images to balance the dataset.
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   What is the primary goal of data gap identification?

   • ATo find missing data elements and plan how to obtain them
   • BTo visualize data distributions for stakeholders
   • CTo encrypt sensitive data fields
   • DTo remove duplicate records from the dataset
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A logistics company is building a model to predict delivery delays. The data science team has access to shipment data, weather data, and traffic data. However, they lack data on road closures, which the business believes is a significant factor. The team is identifying data gaps. What should they do?

   • AUse traffic data as a proxy for road closures.
   • BIdentify the road closure data gap and explore alternative sources, such as government feeds or traffic APIs.
   • CProceed with the available data because road closures are rare and unlikely to affect the model.
   • DAsk the business to remove the road closure requirement from the model's scope.
   Reveal Solution Discussion Select an answer first
5. 14expert · hard

   A manufacturing company is building a predictive maintenance model. The data science team has collected sensor data from 100 machines over one year. However, the target variable (machine failure) is rare, with only 20 failure events. The business wants to deploy the model in a new plant with different machine models. The team must decide whether to proceed. What is the most appropriate decision?

   • AProceed with training but use oversampling techniques to balance the failure events.
   • BDelay the go/no-go decision and conduct a comprehensive data readiness assessment, including failure event sufficiency and representativeness for the new plant.
   • CProceed with training because the sensor data is high-frequency and covers a full year.
   • DProceed with training and later fine-tune the model with data from the new plant.
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A bank is developing a credit risk model. The training data is from the past five years, but the economy has recently entered a recession. The model will be used to evaluate loan applications in the current economic climate. The data science team is validating the data's representativeness. What should they do?

   • ACollect new data from the current recession period before training.
   • BProceed with the historical data because it covers a long time period.
   • CUse the historical data but add a recession indicator variable to the model.
   • DCheck whether the historical data includes periods of economic downturn similar to the current recession.
   Reveal Solution Discussion Select an answer first