AWS AI Practitioner (AIF-C01) · Free practice question 12 of 12
Test set for final unbiased evaluation
A data scientist at Wilmot Analytics splits a dataset into training, validation, and test sets. What is the main purpose of the test set?
- A.To provide a final, unbiased estimate of performance on unseen data after model selection and tuning are complete
- B.To tune hyperparameters repeatedly during training iterations
- C.To supply additional examples for the model to learn from
- D.To store production inference requests for later monitoring
Show answer and explanation
Correct answer: A. To provide a final, unbiased estimate of performance on unseen data after model selection and tuning are complete
Why: The test set is held back until the end and used once to estimate how the chosen model will perform on new, unseen data. The validation set is used during development to tune hyperparameters and compare models, which is why it cannot give an unbiased final estimate. Training data is what the model learns from, and production requests are captured separately for monitoring.
More free AWS AI Practitioner (AIF-C01) questions
- Rekognition image content moderation
- Comprehend Medical clinical text extraction
- Polly text-to-speech narration
- Lex conversational chatbot intents
- Transcribe and Translate call pipeline
- Personalize real-time recommendations
- SageMaker Data Wrangler data preparation
- Deep learning learns features from raw data
- Stop sequences end generation
- SageMaker Feature Store shared features
- SageMaker Model Registry version catalog