1.0 Data Preparation for Machine Learning MLA-C01 Practice Quiz

• A. Apache Parquet is a columnar format efficient for analytics and ML workloads that scan selected columns from large datasets.
• B. JSON is a semi-structured text format used for nested records, APIs, logs, and flexible schema data.
• C. CSV is a simple row-based text format for tabular data that lacks strong typing and efficient compression by default.
• D. Apache ORC is a columnar format optimized for high-performance analytics and efficient storage in big data systems.

Apache Parquet is a columnar format efficient for analytics and ML workloads that scan selected columns from large datasets. This is the correct answer.

• A. Amazon FSx for NetApp ONTAP provides managed NetApp-compatible file storage for workloads that need ONTAP features in AWS.
• B. JSON is a semi-structured text format used for nested records, APIs, logs, and flexible schema data.
• C. Amazon EBS data source provides block storage volumes for EC2-based workloads that need low-latency disk access.
• D. Amazon RDS data source provides managed relational databases that can serve as structured sources for ML preparation workflows.

JSON is a semi-structured text format used for nested records, APIs, logs, and flexible schema data. This is the correct answer.

• A. Amazon Kinesis ingestion is the best fit when an ML data preparation workflow requires this behavior: Amazon Kinesis ingestion provides ingestion and processing of streaming data for near-real-time ML data pipelines.
• B. Apache Parquet is the best fit when an ML data preparation workflow requires this behavior: Apache Parquet is a columnar format efficient for analytics and ML workloads that scan selected columns from large datasets.
• C. Apache Flink streaming is the best fit when an ML data preparation workflow requires this behavior: Apache Flink streaming provides stateful stream processing for real-time transformation and analytics.
• D. Apache Kafka ingestion is the best fit when an ML data preparation workflow requires this behavior: Apache Kafka ingestion provides distributed event streaming for moving data between producers and consumers.

Apache Parquet is a columnar format efficient for analytics and ML workloads that scan selected columns from large datasets. This matches the ML data preparation scenario without confusing data formats, storage services, ingestion mechanisms, transformations, feature engineering, encoding, or data integrity checks. This is the correct answer.

• A. S3 Transfer Acceleration is the best fit when an ML data preparation workflow requires this behavior: S3 Transfer Acceleration provides faster long-distance uploads to S3 by routing traffic through AWS edge locations.
• B. EBS Provisioned IOPS is the best fit when an ML data preparation workflow requires this behavior: EBS Provisioned IOPS provides predictable block storage performance for workloads that require consistent high I/O.
• C. JSON is the best fit when an ML data preparation workflow requires this behavior: JSON is a semi-structured text format used for nested records, APIs, logs, and flexible schema data.
• D. Outlier treatment is the best fit when an ML data preparation workflow requires this behavior: Outlier treatment involves identifying and handling extreme values that may distort model training or analysis.

JSON is a semi-structured text format used for nested records, APIs, logs, and flexible schema data. This matches the ML data preparation scenario without confusing data formats, storage services, ingestion mechanisms, transformations, feature engineering, encoding, or data integrity checks. This is the correct answer.

• A. Normalization is the best fit when an ML data preparation workflow requires this behavior: Normalization involves rescaling data to a defined range so feature magnitudes are comparable.
• B. Feature splitting is the best fit when an ML data preparation workflow requires this behavior: Feature splitting involves separating a compound feature into multiple useful features for modeling.
• C. Binning is the best fit when an ML data preparation workflow requires this behavior: Binning involves grouping continuous numeric values into intervals or categories for modeling or analysis.
• D. CSV is the best fit when an ML data preparation workflow requires this behavior: CSV is a simple row-based text format for tabular data that lacks strong typing and efficient compression by default.

CSV is a simple row-based text format for tabular data that lacks strong typing and efficient compression by default. This matches the ML data preparation scenario without confusing data formats, storage services, ingestion mechanisms, transformations, feature engineering, encoding, or data integrity checks. This is the correct answer.

• A. A workflow points to Deduplication when ingestion, storage, transformation, feature preparation, or data integrity depends on this distinction: Deduplication involves removing duplicate records that could distort training results or evaluation metrics.
• B. A workflow points to Data scaling when ingestion, storage, transformation, feature preparation, or data integrity depends on this distinction: Data scaling involves adjusting numeric feature ranges so algorithms sensitive to magnitude can train more effectively.
• C. A workflow points to Apache Parquet when ingestion, storage, transformation, feature preparation, or data integrity depends on this distinction: Apache Parquet is a columnar format efficient for analytics and ML workloads that scan selected columns from large datasets.
• D. A workflow points to Standardization when ingestion, storage, transformation, feature preparation, or data integrity depends on this distinction: Standardization involves transforming numeric features to a common scale, often centered around a mean with a standard deviation.

Apache Parquet is a columnar format efficient for analytics and ML workloads that scan selected columns from large datasets. This distinction matters because choosing a nearby concept would lead to the wrong data format, storage service, ingestion path, transformation, feature representation, or quality-control step. This is the correct answer.

• A. A workflow points to One-hot encoding when ingestion, storage, transformation, feature preparation, or data integrity depends on this distinction: One-hot encoding involves converting categorical values into separate binary columns.
• B. A workflow points to Label encoding when ingestion, storage, transformation, feature preparation, or data integrity depends on this distinction: Label encoding involves converting categorical values into numeric labels for algorithms that can interpret encoded categories appropriately.
• C. A workflow points to Tokenization when ingestion, storage, transformation, feature preparation, or data integrity depends on this distinction: Tokenization involves splitting text into units such as words, subwords, or tokens for NLP workflows.
• D. A workflow points to JSON when ingestion, storage, transformation, feature preparation, or data integrity depends on this distinction: JSON is a semi-structured text format used for nested records, APIs, logs, and flexible schema data.

JSON is a semi-structured text format used for nested records, APIs, logs, and flexible schema data. This distinction matters because choosing a nearby concept would lead to the wrong data format, storage service, ingestion path, transformation, feature representation, or quality-control step. This is the correct answer.

• A. Data quality validation is the correct ML engineering choice when the data preparation decision depends on this exact behavior: Data quality validation involves checking data for completeness, accuracy, consistency, schema correctness, and suitability for modeling.
• B. JSON is the correct ML engineering choice when the data preparation decision depends on this exact behavior: JSON is a semi-structured text format used for nested records, APIs, logs, and flexible schema data.
• C. CSV is the correct ML engineering choice when the data preparation decision depends on this exact behavior: CSV is a simple row-based text format for tabular data that lacks strong typing and efficient compression by default.
• D. Apache Parquet is the correct ML engineering choice when the data preparation decision depends on this exact behavior: Apache Parquet is a columnar format efficient for analytics and ML workloads that scan selected columns from large datasets.

Apache Parquet is a columnar format efficient for analytics and ML workloads that scan selected columns from large datasets. This applies the concept at the point where an ML engineer must prepare data for modeling while balancing access pattern, cost, performance, quality, and modeling readiness. This is the correct answer.