Dimensionality reduction and data sampling: Practice Questions — Data Manipulation and Preparation (NVIDIA-Certified Associate: Accelerated Data Science)
Practice Questions: Dimensionality Reduction and Data Sampling These multiple-choice questions are designed to help you prepare for the...
Practice Questions: Dimensionality Reduction and Data Sampling
These multiple-choice questions are designed to help you prepare for the Dimensionality Reduction and Data Sampling portion of the NVIDIA-Certified Associate: Accelerated Data Science exam. Each question includes four options, the correct answer, and a brief explanation.
Which of the following techniques is primarily used to reduce the number of features while preserving the variance in the dataset?
- A) Random Sampling
- B) Principal Component Analysis (PCA)
- C) SMOTE (Synthetic Minority Over-sampling Technique)
- D) One-Hot Encoding
Answer: B) Principal Component Analysis (PCA)
Explanation: PCA transforms the original features into a smaller set of uncorrelated components that capture the most variance, effectively reducing dimensionality.
When performing data sampling to handle large datasets, which approach ensures that the sample maintains the original class distribution?
- A) Stratified Sampling
- B) Simple Random Sampling
- C) Systematic Sampling
- D) Cluster Sampling
Answer: A) Stratified Sampling
Explanation: Stratified sampling divides the dataset into strata based on class labels and samples proportionally, preserving class distribution.
Which dimensionality reduction method is best suited for nonlinear relationships in data?
- A) Linear Discriminant Analysis (LDA)
- B) t-Distributed Stochastic Neighbor Embedding (t-SNE)
- C) Feature Scaling
- D) Min-Max Normalization
Answer: B) t-Distributed Stochastic Neighbor Embedding (t-SNE)
Explanation: t-SNE is a nonlinear technique that preserves local structure and is effective for visualizing high-dimensional data with complex relationships.
In GPU-accelerated data science workflows, which library is commonly used for efficient in-memory data sampling and manipulation?
- A) cuDF
- B) TensorFlow
- C) Matplotlib
- D) Scikit-learn
Answer: A) cuDF
Explanation: cuDF is a RAPIDS library that provides GPU-accelerated DataFrame operations, including efficient data sampling and manipulation.
What is the main purpose of dimensionality reduction before training machine learning models?
- A) Increase the number of features to improve accuracy
- B) Reduce computational cost and mitigate overfitting
- C) Remove all categorical variables
- D) Generate synthetic data
Answer: B) Reduce computational cost and mitigate overfitting
Explanation: Reducing dimensionality simplifies models, lowers training time, and helps prevent overfitting by eliminating redundant or irrelevant features.
Which of the following best describes the role of data sampling in the context of GPU-accelerated ETL pipelines?
- A) To increase dataset size for better model training
- B) To reduce dataset size for faster processing without losing representativeness
- C) To convert categorical variables into numerical ones
- D) To generate synthetic features
Answer: B) To reduce dataset size for faster processing without losing representativeness
Explanation: Data sampling reduces the volume of data processed, enabling faster GPU-accelerated ETL while maintaining the integrity of the dataset.
Which technique is used to reduce dimensionality by selecting a subset of original features rather than creating new combinations?
- A) Feature Extraction
- B) Feature Selection
- C) PCA
- D) SMOTE
Answer: B) Feature Selection
Explanation: Feature selection chooses the most relevant original features, whereas feature extraction (like PCA) creates new features by combining existing ones.
In the context of NVIDIA RAPIDS, which tool can be used to perform distributed data sampling on large datasets?
- A) Dask
- B) Pandas
- C) Matplotlib
- D) NumPy
Answer: A) Dask
Explanation: Dask enables scalable, distributed data processing and sampling on large datasets, integrating seamlessly with RAPIDS GPU-accelerated libraries.
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 →