MLS-C01 Practice Test 41 — 15 Questions

Question 1

A company uses Amazon SageMaker to train a deep learning model for image classification. The training job is taking longer than expected. The data scientist observes that GPU utilization is low (around 30%) and CPU utilization is high. Which action is most likely to reduce training time?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Increase the number of data loading workers. Low GPU utilization with high CPU utilization indicates a data loading bottleneck where the CPU cannot prepare batches fast enough to keep the GPU busy. Increasing the number of data loading workers (e.g., SageMaker's `sagemaker.session.Session` or PyTorch `DataLoader` `num_workers`) allows parallel data preprocessing and I/O, reducing idle GPU time and overall training duration.

Answer

Reduce the batch size

Answer

Increase the batch size

Answer

Increase the learning rate

Question 2

A data scientist is analyzing a time series dataset of daily website traffic. The scientist notices a strong weekly seasonality. To better understand the underlying patterns, which decomposition method should the scientist use to separate the trend, seasonal, and residual components?

Accepted Answer

Use STL (Seasonal and Trend decomposition using Loess).. STL (Seasonal and Trend decomposition using Loess) is a robust method for decomposing time series into trend, seasonal, and residual components. It can handle any seasonality period, including weekly seasonality in daily data, and is robust to outliers. Option A is wrong because additive decomposition using moving averages assumes fixed seasonal amplitude and is sensitive to outliers. Option C is wrong because ARIMA is a forecasting model, not a decomposition method. Option D is wrong because ETS is an exponential smoothing framework for forecasting, not primarily for decomposition.

Answer

Additive decomposition using moving averages.

Answer

Fit an ARIMA model and examine residuals.

Answer

Apply an ETS (Error, Trend, Seasonal) model.

Question 3

Which TWO approaches are appropriate for handling missing categorical data during exploratory data analysis? (Choose two.)

Accepted Answer

Impute with the mode (most frequent) of the column.. Option B is correct because imputing missing categorical values with the mode (most frequent category) preserves the column's most common label and keeps the dataset size intact, which is a standard and appropriate EDA technique for categorical variables. Option C is correct because treating missing values as a distinct 'Unknown' category retains the information that the value was missing and lets the analysis capture any pattern associated with missingness, which is often meaningful in categorical data. Option A is not appropriate here because one-hot encoding is a representation technique for existing categories, not a missing-data handling strategy, and encoding missingness as a binary feature alone does not resolve the missing categorical value. Option D is not appropriate because dropping all rows with missing values discards potentially large amounts of data and can introduce bias, which is generally undesirable during EDA. Option E is not appropriate because the mean is only defined for numeric data and is meaningless for categorical variables.

Answer

Use one-hot encoding to represent missingness as a binary feature.

Answer

Drop all rows with missing values in that column.

Answer

Impute missing values with the mean of the column.

Question 4

An e-commerce company uses Amazon Redshift for analytics. The data engineering team needs to load daily sales data from an S3 bucket that receives new files every hour. The data must be loaded into Redshift with minimal impact on query performance during the day, and they need to handle late-arriving data (files that appear after the daily load). Which approach should they use?

Accepted Answer

Use a staging table to load data incrementally with a MERGE operation, and schedule a late-arriving data job to merge files that arrive after the daily load.. It uses a staging table to incrementally load data with a MERGE operation, which minimizes impact on query performance by avoiding full table overwrites. The separate late-arriving data job handles files that appear after the daily load, ensuring completeness without blocking ongoing queries. This approach aligns with Redshift's best practices for incremental loads and late-arriving data handling.

Answer

Use AWS Glue ETL to copy the data from S3 to Redshift, overwriting the existing data each day.

Answer

Stream the data from S3 using Amazon Kinesis Firehose to load into Redshift continuously.

Answer

Use Amazon Redshift Spectrum to query data directly from S3 and create external tables.

Question 5

A machine learning engineer is deploying a model using Amazon SageMaker. The model is a PyTorch model that performs real-time inference with low latency requirements. The engineer wants to use automatic scaling based on the number of concurrent requests. Which SageMaker feature should be used to achieve this?

Accepted Answer

Use Amazon SageMaker automatic scaling with a target tracking scaling policy.. Amazon SageMaker automatic scaling with a target tracking scaling policy is the correct feature because it allows the endpoint to dynamically adjust the number of instances based on a predefined metric, such as the number of concurrent requests (e.g., using the SageMakerVariantInvocationsPerInstance metric). This directly meets the requirement for automatic scaling based on concurrent requests while maintaining low latency for real-time PyTorch inference.

Answer

Create an AWS Auto Scaling group for the SageMaker endpoint.

Answer

Enable Elastic Load Balancing for the endpoint.

Answer

Deploy the model behind Amazon API Gateway with a Lambda function.

Question 6

During exploratory data analysis, a data scientist plots the distribution of a numerical feature and observes a heavy right skew. The feature has many outliers at the high end. Which transformation is most appropriate to reduce skewness?

Accepted Answer

Apply a log transformation to the feature.. A log transformation compresses the range of the data, reducing the impact of extreme values and pulling in the long tail of a right-skewed distribution. This makes the feature more normally distributed, which is often required for linear models and many statistical tests. It is the standard technique for handling positive-valued features with heavy right skew.

Answer

Apply z-score normalization.

Answer

Apply one-hot encoding.

Answer

Apply min-max scaling.

Question 7

A company is training a deep learning model on Amazon SageMaker. The training job is taking a long time and the data scientist suspects that the model is overfitting. Which of the following actions can help reduce overfitting and improve generalization?

Accepted Answer

Add dropout layers to the model architecture.. Dropout is a regularization technique that randomly deactivates a fraction of neurons during each training step, forcing the network to learn redundant, distributed representations rather than memorizing training samples. This directly combats overfitting by reducing co-adaptation of neurons and effectively approximating an ensemble of subnetworks at inference time. In SageMaker training jobs, adding dropout layers (e.g., via the framework's model definition) is a standard, low-cost architectural change that improves generalization without altering the data pipeline.

Answer

Increase the batch size used during training.

Answer

Increase the number of training epochs.

Answer

Remove regularization terms from the loss function.

Question 8

A data scientist runs a SQL query on an Amazon Athena table and notices that the query scans a large amount of data. Which approach would reduce the amount of data scanned without changing the SQL logic?

Accepted Answer

Partition the table on a column that is frequently used in WHERE clauses.. Partitioning an Athena table on a column frequently used in WHERE clauses allows Athena to prune partitions and read only the relevant data, drastically reducing the amount of data scanned. This is because Athena uses the partition metadata in the AWS Glue Data Catalog to skip entire partitions that do not match the query predicate. Since the SQL logic remains unchanged, the query still returns the same results but scans less data, lowering cost and improving performance.

Answer

Convert the data from CSV to JSON format.

Answer

Store the data in Parquet format without partitioning.

Answer

Use GZIP compression on the data files.

Question 9

A deployed SageMaker endpoint is returning high latency. The model is a scikit-learn Random Forest. Which action is most likely to reduce latency?

Accepted Answer

Reduce the number of trees in the ensemble. Reducing the number of trees in a Random Forest ensemble directly decreases the total number of decision paths that must be evaluated per inference request. Since each tree contributes additively to the prediction time, fewer trees means fewer sequential or parallel evaluations, which lowers the per-request latency at the cost of some model accuracy.

Answer

Prune decision trees in the model

Answer

Increase the number of instances behind the endpoint

Answer

Switch to a GPU instance type

Question 10

A machine learning team is preparing a large dataset for training. The dataset consists of 10,000 CSV files, each about 100 MB, stored in Amazon S3. The team wants to transform the data using AWS Glue ETL jobs. The transformation involves filtering rows, adding new columns, and joining with a small reference table (100 KB). The team is concerned about job performance and cost. They currently have a Glue job with 10 DPU (Data Processing Units) and it takes about 2 hours to complete. The team wants to reduce the runtime and cost. Which approach should they take?

Accepted Answer

Convert the CSV files to Parquet format and partition the data by a column.. Converting the CSV files to Parquet format and partitioning the data by a column significantly reduces the amount of data scanned and processed by AWS Glue. Parquet is a columnar storage format that allows Glue to read only the necessary columns, and partitioning enables predicate pushdown to skip irrelevant partitions. This directly reduces I/O and compute requirements, leading to faster job runtime and lower cost without increasing DPU count.

Answer

Use Amazon Athena to transform the data.

Answer

Increase the number of DPUs to 100.

Answer

Use Amazon EMR with Spot Instances instead of AWS Glue.

Question 11

A data scientist is performing exploratory data analysis on a high-dimensional dataset with 500 features. The scientist wants to visualize the data in 2D to check for clusters. Which dimensionality reduction technique should the scientist use that preserves global structure and is computationally efficient for large datasets?

Accepted Answer

PCA. PCA is a linear dimensionality reduction technique that preserves global structure (variance) and is computationally efficient for large datasets. Option A is wrong because t-SNE is non-linear, slower, and focuses on preserving local structure, not global. Option B is wrong because LDA is a supervised technique that requires class labels, and it is not typically used for unsupervised exploration of clusters. Option D is wrong because UMAP is non-linear and can be slower than PCA; while it preserves both local and global structure to some extent, it is not as computationally efficient as PCA for very large datasets.

Answer

t-SNE

Answer

Linear Discriminant Analysis (LDA)

Answer

UMAP

Question 12

A company stores sensor data in Amazon S3. A data scientist wants to explore the data using SQL without moving it. Which AWS service should they use?

Accepted Answer

Amazon Athena. Amazon Athena is the correct choice because it is a serverless interactive query service that allows you to analyze data directly in Amazon S3 using standard SQL without any data movement or infrastructure management. Athena uses Presto under the hood and charges only for the data scanned per query, making it ideal for ad-hoc exploratory analysis on sensor data stored in S3.

Answer

Amazon EMR

Answer

Amazon Redshift

Answer

Amazon QuickSight

Question 13

A data scientist is building a model to predict customer churn. The dataset contains categorical features with high cardinality (e.g., ZIP code, customer ID). Which encoding method is MOST suitable?

Accepted Answer

Target encoding. Target encoding is most suitable for high-cardinality categorical features because it replaces each category with the mean of the target variable for that category, effectively capturing the predictive signal while keeping the feature space dense. This avoids the curse of dimensionality from one-hot encoding and the arbitrary ordinality of label encoding, which can mislead tree-based models.

Answer

One-hot encoding

Answer

Label encoding

Answer

Hashing encoding

Question 14

Which TWO statements about handling categorical variables in exploratory data analysis are correct? (Select TWO.)

Accepted Answer

When a categorical feature has high cardinality, consider grouping rare categories.. High-cardinality categorical features can lead to overfitting and sparse representations. Grouping rare categories into a single 'Other' bucket reduces dimensionality and noise, improving model generalization without losing significant predictive signal.

Answer

Target encoding always avoids data leakage.

Answer

Label encoding is suitable for nominal categorical variables.

Answer

Categorical variables should always be dropped if they have many unique values.

Question 15

A data scientist is using Amazon SageMaker to train a model with the built-in XGBoost algorithm. The dataset contains missing values. What is the default behavior of SageMaker XGBoost regarding missing values?

Accepted Answer

It automatically learns the best direction (left or right) for missing values during training. SageMaker's built-in XGBoost algorithm handles missing values natively by learning a default direction for each tree split: during training, when a feature value is missing, the algorithm assigns it to the left or right branch based on which direction minimizes loss. This is the same sparse-aware behavior as open-source XGBoost, and it means no explicit imputation is required. The 'missing' parameter in SageMaker XGBoost defaults to 'NaN', enabling this automatic handling.

Answer

It raises an error and stops training

Answer

It imputes missing values with the column mean

Answer

It removes rows with missing values