Objective 4.3 Evaluate Models CERTIFIED-DATA-SCIENCE-PRACTITIONER Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   A machine learning team needs to automate a multi-step workflow that includes data preprocessing, model training, and model deployment to a Kubernetes cluster. They want to version the pipeline and rerun it when new data arrives. Which tool is most appropriate?

   • AApache Airflow
   • BDocker Compose
   • CKubeflow Pipelines
   • DMLflow Tracking
   Reveal Solution Discussion Select an answer first
2. 2foundation · easy

   What is the primary purpose of an evaluation metric in machine learning?

   • ATo automatically improve the model's accuracy during training
   • BTo store the model in a deployment-ready format
   • CTo quantify how well a model performs on a given task
   • DTo visualize the model's architecture and parameters
   Reveal Solution Discussion Select an answer first
4. 3expert · hard

   A team has a Kubeflow pipeline that trains a model and deploys it to a Kubernetes cluster. They need to update the model weekly with new data and ensure that the deployment is rolled back automatically if the new model's accuracy drops below a threshold. Which Kubeflow component is best suited for this?

   • AKubeflow Fairing
   • BKubeflow Katib
   • CKubeflow Pipelines with a condition on the model's accuracy metric to trigger deployment or rollback.
   • DKubeflow Serving
   Reveal Solution Discussion Select an answer first
5. 4expert · hard

   A team is training a deep neural network for image classification. The learning curve shows that both training and validation error are high and nearly identical. They have already tried increasing model capacity and adding more data, but the curves remain flat. What is the most likely issue and the best next step?

   • AIrrelevant features; perform feature selection.
   • BData leakage; re-split the data.
   • CHigh variance; apply more regularization.
   • DHigh bias; try a more complex model architecture or better feature engineering.
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   If a learning curve shows a large gap between training and validation scores, with training score high and validation score low, what issue is indicated?

   • AData leakage
   • BHigh variance (overfitting)
   • CNo issue; this is normal
   • DHigh bias (underfitting)
   Reveal Solution Discussion Select an answer first