A data scientist is preprocessing a dataset containing multiple categorical features using NVIDIA RAPIDS to accelerate feature engineering.
The dataset contains:
A low-cardinality categorical feature (Product Type) with 10 unique values.
A high-cardinality categorical feature (User ID) with 100,000 unique values.
A numerical feature (Price) that requires transformation.
Which of the following feature engineering approaches will be the most efficient for GPU acceleration?

• A. Convert both Product Type and User ID to int64 and use standardization (mean normalization) on Price.
• B. Apply one-hot encoding to both Product Type and User ID, and scale Price using float64 precision.
• C. Frequency encoding for User ID is an efficient alternative to one-hot encoding, as it replaces each category with its frequency in the dataset, reducing dimensionality while preserving useful information.
• D. Using float32 for Price is optimal for GPU-based ML models, balancing precision and computational efficiency.
• E. Store both Product Type and User ID as string data types in cuDF to maintain raw categorical information.
• F. Convert Product Type to integers using label encoding, use frequency encoding for User ID, and normalize Price using float32.

A financial institution is developing an ETL pipeline to ingest and process large volumes of streaming data from various sources, including stock market feeds, real-time transactions, and economic indicators. The ETL process must be highly efficient to minimize latency while ensuring data integrity.
Which of the following strategies is best suited for implementing a high-performance, GPU-accelerated ETL pipeline?

• A. Store all streaming data in a PostgreSQL database before performing batch transformations.
• B. Load data directly into an Excel spreadsheet and use VBA macros to clean and transform it.
• C. Use Pandas and Python's built-in threading library to handle concurrent data ingestion and transformation.
• D. Utilize NVIDIA Morpheus with RAPIDS to preprocess real-time streaming data using GPU acceleration.

A data scientist is working with a 50 TB dataset consisting of structured logs from IoT devices. The data needs to be cleaned, transformed, and aggregated before training a machine learning model.
Which of the following frameworks would be the most efficient choice for distributed data processing?

• A. SQLite
• B. Pandas
• C. Apache Spark with RAPIDS Accelerator
• D. Python multiprocessing module

Which of the following is the main advantage of using TensorRT for inference in an accelerated data science pipeline?

• A. TensorRT optimizes deep learning models to run efficiently on NVIDIA GPUs by reducing precision while maintaining accuracy.
• B. TensorRT automatically builds training models from raw data without requiring pre-trained models.
• C. TensorRT is only compatible with image classification models and does not support other model types.
• D. TensorRT is mainly used for data visualization and not for model inference.