Feature engineering, selection, and transformation: Practice Questions — Data Science Pipelines and Workflow Automation (NVIDIA-Certified Associate: Accelerated Data Science)
Practice Questions: Feature Engineering, Selection, and Transformation These multiple-choice questions are designed to help candidates prepare for...
Practice Questions: Feature Engineering, Selection, and Transformation
These multiple-choice questions are designed to help candidates prepare for the NVIDIA-Certified Associate: Accelerated Data Science exam, focusing on the critical subtopic of feature engineering, selection, and transformation within data science pipelines.
Which of the following best describes feature scaling and why is it important in machine learning pipelines?
- A. Encoding categorical variables into numerical values to enable model training
- B. Adjusting the range of feature values to a standard scale to improve model convergence
- C. Removing irrelevant features to reduce model complexity
- D. Creating new features by combining existing ones to capture interactions
Correct answer: B
Explanation: Feature scaling adjusts numerical feature values to a common scale (e.g., normalization or standardization), which helps many machine learning algorithms converge faster and perform better.
What is the primary goal of feature selection in a data science pipeline?
- A. To increase the dimensionality of the dataset for better model expressiveness
- B. To identify and retain the most relevant features while removing redundant or noisy ones
- C. To transform features into a different space using mathematical functions
- D. To generate synthetic data points to augment the dataset
Correct answer: B
Explanation: Feature selection aims to improve model performance and reduce overfitting by keeping only the most informative features and discarding irrelevant or redundant ones.
Which technique is commonly used for feature transformation to reduce feature correlation and dimensionality?
- A. One-hot encoding
- B. Principal Component Analysis (PCA)
- C. Min-max scaling
- D. Data augmentation
Correct answer: B
Explanation: PCA transforms correlated features into a set of linearly uncorrelated components, reducing dimensionality while preserving variance.
In the context of mitigating overfitting, which feature engineering practice is most effective?
- A. Adding more irrelevant features
- B. Using feature selection to remove noisy or redundant features
- C. Increasing the number of features without validation
- D. Avoiding any feature transformation
Correct answer: B
Explanation: Removing noisy or irrelevant features helps reduce model complexity and overfitting, improving generalization.
Which of the following is an example of feature augmentation?
- A. Scaling numerical features to a standard range
- B. Creating polynomial features by combining existing numerical features
- C. Selecting the top k features based on importance scores
- D. Encoding categorical variables using label encoding
Correct answer: B
Explanation: Feature augmentation involves creating new features, such as polynomial combinations, to capture complex relationships in the data.
When building reproducible pipelines with RAPIDS and Dask, why is it important to automate feature engineering steps?
- A. To ensure consistent preprocessing across different datasets and runs
- B. To manually tune each feature transformation for every experiment
- C. To avoid using GPU acceleration for feature processing
- D. To increase the risk of data leakage
Correct answer: A
Explanation: Automating feature engineering ensures that the same transformations are applied consistently, which is critical for reproducibility and reliable model evaluation.
Which method is most appropriate for encoding high-cardinality categorical features in a GPU-accelerated pipeline?
- A. One-hot encoding
- B. Label encoding
- C. Target encoding
- D. Dropping the feature
Correct answer: C
Explanation: Target encoding replaces categories with a target-based statistic, reducing dimensionality compared to one-hot encoding, which is beneficial for high-cardinality features in GPU-accelerated workflows.
What is the main advantage of using automated feature selection techniques in accelerated data science pipelines?
- A. They eliminate the need for any human oversight
- B. They speed up model training by reducing feature set size while maintaining performance
- C. They always select the features with the highest correlation to the target
- D. They increase the complexity of the pipeline
Correct answer: B
Explanation: Automated feature selection methods help efficiently reduce the number of features, which speeds up training and can improve model generalization without extensive manual effort.
More in this topic
Ready to test your knowledge?
Put what you've learned into practice with a quick quiz and track your progress.
Test your knowledge →