Operationalize and Monitor Models SNOWPRO-ADVANCED-DATA-SCIENTIST Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   A software company has a model that predicts software defects. The model is updated regularly, and each version is stored in a model registry. After deploying a new version, the team notices that the model's false positive rate has increased significantly. They want to roll back to the previous version. What is the best practice for executing the rollback?

   • AImmediately delete the new model version from the registry to prevent further use.
   • BRedeploy the previous model version and keep the new version in the registry for analysis.
   • CKeep the new model and adjust the decision threshold to reduce false positives.
   • DUpdate the current model's code to fix the issue and redeploy it.
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   An insurance company has a model that predicts claim fraud. The model is monitored for data drift. Recently, the data science team noticed that the distribution of claim amounts has shifted, but the model's accuracy has not changed. What is the most appropriate interpretation and action?

   • AThe drift is a false positive, so it can be ignored.
   • BThe model is still performing well, so no action is needed.
   • CThe model should be retrained immediately to prevent future degradation.
   • DThe drift may not have impacted accuracy yet, but it should be monitored closely and the model may need retraining soon.
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   Which of the following is typically included in an operational dashboard for model monitoring?

   • AThe model's hyperparameter values
   • BModel prediction latency over time
   • CThe raw training data
   • DThe source code of the model
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   What is the main purpose of an operational dashboard for a deployed ML model?

   • ATo automatically retrain the model when performance drops
   • BTo store the model's training data
   • CTo replace the need for model versioning
   • DTo provide a real-time view of model health, usage, and performance for stakeholders
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   Which metric specifically measures the degree to which the distribution of model predictions has shifted from the distribution observed during training?

   • APrediction drift
   • BPrediction accuracy
   • CModel latency
   • DData drift
   Reveal Solution Discussion Select an answer first