MLS-C01 Mock Exam 75 — 25 Questions

Question 1

A data scientist is using Amazon SageMaker to train a model and needs to monitor the training job's resource utilization and detect anomalies such as high CPU usage or memory leaks. The data scientist wants to receive alerts when the training job's CPU utilization exceeds 90% for more than 5 minutes. Which AWS service should be used to set up these alerts?

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

Accepted Answer

Amazon CloudWatch Alarms with the CPUUtilization metric for the training job.. Amazon CloudWatch is the AWS service for monitoring metrics and setting alarms. SageMaker training jobs automatically publish metrics such as CPUUtilization to CloudWatch. By creating a CloudWatch Alarm on the CPUUtilization metric with a threshold and period, the data scientist can receive alerts when CPU usage exceeds 90% for 5 minutes. Other services like CloudTrail, Debugger, and Trusted Advisor do not provide this specific alerting capability.

Answer

AWS CloudTrail with a trail that logs SageMaker API calls and filters for high CPU events.

Answer

Amazon SageMaker Debugger with built-in rules to detect high CPU utilization.

Answer

AWS Trusted Advisor with a check for SageMaker training job CPU usage.

Question 2

A data scientist is exploring a dataset where two continuous features have a strong monotonic but clearly nonlinear relationship. Pearson correlation between them is near zero, yet the scatter plot shows a consistent increasing curve. The scientist wants a single numeric summary that quantifies the strength of this monotonic association. Which statistic should the scientist compute?

Accepted Answer

Spearman rank correlation coefficient between the two features.. A near-zero Pearson correlation combined with an obvious increasing curve signals a monotonic but nonlinear relationship. Rank-based correlation measures association through ordering rather than linear fit, so it captures the curve as a strong positive value. The other statistics either measure only linear relationships or are designed for categorical data, making them unsuitable for summarizing the monotonic association the scientist observed.

Answer

Chi-square test statistic computed on binned versions of the two features.

Answer

Pearson correlation coefficient computed on the raw feature values.

Answer

Covariance between the two features after subtracting their means.

Question 3

An ML engineer is preparing a SageMaker training job that reads a large dataset from Amazon S3. The dataset consists of many Parquet files under a single S3 prefix, and the job will run on a multi-instance cluster. The engineer wants to reduce training startup time and maximize read throughput across instances. (Choose two.)

Accepted Answer

Set the input channel's S3DataDistributionType to ShardedByS3Key so each instance reads a distinct subset of the objects.. ShardedByS3Key spreads the Parquet objects across the instances so each reads a distinct subset, and Pipe mode streams data directly from S3 instead of staging everything locally first. Together they reduce redundant transfer and startup delay while raising aggregate read throughput on a multi-instance training cluster.

Answer

Set the input channel's S3DataDistributionType to FullyReplicated so every instance has a complete copy of the dataset.

Answer

Increase the training job's MaxRuntimeInSeconds and MaxWaitTimeInSeconds parameters to allow more time for data download.

Answer

Enable Amazon S3 Transfer Acceleration on the bucket to speed up downloads to the training instances.

Question 4

A data analyst has a CSV file of customer transactions in Amazon S3 and needs a quick, visual profile of each column: inferred data types, missing-value counts, value ranges, and a correlation matrix, without writing any transformation code. Which AWS capability should the analyst use?

Accepted Answer

Run the AWS Glue DataBrew profile job on the dataset and review the generated profile in the console.. AWS Glue DataBrew's profile job automatically infers column types and computes summary statistics, missing-value counts, ranges, and correlations, all through a visual interface over S3 data. That matches the requirement for a fast, code-free column profile of the transaction CSV without building transformations.

Answer

Enable Amazon Macie on the S3 bucket and inspect the sensitive data findings.

Answer

Attach an AWS Glue crawler to the prefix and view the table schema in the Glue Data Catalog.

Answer

Create an Amazon QuickSight dataset and add a scatter plot of two numeric columns.

Question 5

A data analyst has a 5 GB CSV dataset in Amazon S3 and wants to visualize feature distributions, missing-value patterns, and pairwise scatter plots during EDA without writing code. She has no existing AWS Glue Data Catalog tables. Which AWS service should she use?

Accepted Answer

Amazon SageMaker Data Wrangler, importing the CSV directly from S3 into a data flow and using its built-in analysis visualizations.. Data Wrangler is purpose-built for no-code exploratory analysis: it imports CSV directly from S3 without a Data Catalog dependency, and its analysis view renders histograms and a missing-value heatmap. The other services either require schema registration first, are dashboard-oriented, or add a separate profiling job lifecycle, so they do not match the direct, visual, code-free requirement.

Answer

Amazon Athena, creating an external table over the S3 prefix and running SELECT queries to inspect distributions.

Answer

AWS Glue crawler plus the Glue DataBrew console, running a profile job on the crawled table.

Answer

Amazon QuickSight, connecting to the S3 bucket as a data source and building analysis dashboards.

Question 6

A machine learning engineer is using Amazon SageMaker to train a model for predicting housing prices. The dataset contains 20,000 records with 15 numerical features. The engineer notices that the model's predictions are consistently off by a large margin for high-priced houses. Which SageMaker built-in algorithm is BEST suited for this regression task?

Accepted Answer

XGBoost. For regression on tabular data with potential nonlinear relationships, XGBoost is a strong choice. It can model complex interactions and is robust to outliers. Linear Regression may underfit, while K-Means and DeepAR are for unsupervised and time-series tasks, respectively.

Answer

DeepAR

Answer

K-Means

Answer

Linear Regression

Question 7

A data scientist is exploring a dataset with a highly skewed numerical feature representing customer income. The feature has a long right tail and several extreme outliers. The scientist wants to apply a transformation that reduces skewness and makes the distribution more symmetric for subsequent modeling. Which transformation is MOST appropriate?

Accepted Answer

Logarithmic transformation (log(x + 1)). Logarithmic transformation is a variance-stabilizing technique that compresses the upper tail, reducing right skewness and mitigating the influence of outliers. It is well-suited for positive, skewed data such as income. Standardization and min-max scaling only change location and scale, not shape, while binning discards information and does not create a symmetric continuous distribution.

Answer

Min-max scaling to [0, 1]

Answer

Binning into equal-width bins

Answer

Standardization (z-score normalization)

Question 8

A company uses Amazon SageMaker to train models on data stored in Amazon S3. The data is in CSV format and is updated daily. The training jobs need to read the entire dataset for each run. The data engineering team wants to minimize the time spent downloading data to the training instances and reduce the overall training time. Which solution should they implement?

Accepted Answer

Convert the CSV data to RecordIO format and use Pipe mode in the SageMaker training job.. The goal is to minimize data download time and reduce overall training time. Pipe mode streams data directly from S3 to the training algorithm, eliminating the need to download the full dataset. Converting CSV to RecordIO improves streaming efficiency and is compatible with SageMaker built-in algorithms. This combination reduces training time and disk usage. Other options either still download the data or add unnecessary complexity.

Answer

Use SageMaker's FastFile mode with the CSV data to stream data on demand.

Answer

Use Amazon FSx for Lustre as the data source for the SageMaker training job.

Answer

Use File mode with the CSV data and increase the size of the training instance's EBS volume.

Question 9

A company runs an AWS Glue ETL job that processes data incrementally. The job uses job bookmarks to track previously processed S3 objects. After a schema change in the source data, the team notices that newly arriving objects are not being processed. They confirm the job bookmark is enabled and that the new objects have different ETags and modification times. Which action should the data engineer take?

Accepted Answer

Reset the job bookmark and rerun the job so it reprocesses the source from the beginning with the updated schema.. Glue job bookmarks store state about which objects and partitions were previously processed. A source schema change can leave the bookmark unable to recognize new objects as unprocessed. Resetting the bookmark clears that state so the next run reprocesses the source with the updated schema, restoring processing of the newly arriving objects.

Answer

Convert the job to use AWS Glue DataBrew instead, because DataBrew automatically handles schema changes with bookmarks.

Answer

Disable job bookmarks temporarily and enable them again after the next successful run to refresh the bookmark state.

Answer

Increase the Glue job's number of workers and enable auto scaling to give the bookmark logic more resources to detect new objects.

Question 10

A data scientist is training a binary classification model using Amazon SageMaker's built-in Linear Learner algorithm. The dataset is highly imbalanced, with only 2% of the samples belonging to the positive class. The data scientist wants to adjust the algorithm to better handle the imbalance. Which hyperparameter should be modified?

Accepted Answer

Set the positive_example_weight_mult parameter to a value greater than 1.. The SageMaker Linear Learner algorithm provides the positive_example_weight_mult hyperparameter to handle imbalanced datasets by weighting the positive class more heavily. This is the correct approach because it directly influences the loss function to penalize misclassifications of the minority class more. The other options reference non-existent parameters or do not effectively address the imbalance.

Answer

Set the target_recall parameter to a high value.

Answer

Set the loss parameter to 'logistic' and increase the learning rate.

Answer

Set the balance_multiplier parameter to a value less than 1.

Question 11

A company has deployed a SageMaker endpoint for a fraud detection model. The model was trained on historical data and is performing well. However, the company notices that the model's predictions are becoming less accurate over time due to changes in customer behavior. The company wants to detect this degradation and receive alerts when the model's performance drops below a threshold. Which SageMaker feature should the company use?

Accepted Answer

SageMaker Model Monitor with a baseline and monitoring schedule. SageMaker Model Monitor is designed to continuously monitor the quality of predictions from a deployed model. By setting up a baseline and a monitoring schedule, it can detect data drift and model quality issues, and trigger CloudWatch alarms when violations exceed a threshold. This enables the company to be alerted when model performance degrades due to changing customer behavior.

Answer

SageMaker Clarify

Answer

SageMaker Inference Recommender

Answer

SageMaker Debugger with built-in rules

Question 12

A data scientist is exploring a dataset with a highly skewed numerical feature that has a long right tail. The scientist wants to visualize the distribution to decide on a transformation. Which visualization technique is BEST suited for this purpose?

Accepted Answer

A histogram with a logarithmic scale on the x-axis.. A histogram with a logarithmic x-axis is ideal for visualizing a highly skewed feature because it compresses the long tail and reveals the underlying distribution shape. This helps determine if a log transformation is appropriate and provides a clearer view of the data's spread and central tendency.

Answer

A box plot of the feature grouped by a categorical variable.

Answer

A line plot of the feature sorted by value.

Answer

A scatter plot of the feature against another numerical feature.

Question 13

A data engineer must load a 5 GB uncompressed CSV file from Amazon S3 into Amazon Redshift using the COPY command. The cluster has four nodes and the file resides in a bucket in the same AWS Region. The load is taking far longer than expected and the engineer notices the data is being distributed unevenly across slices. What is the most likely cause?

Accepted Answer

The COPY command is reading a single file, so only one slice is used to parse and load the data.. COPY achieves parallelism by distributing input files across the cluster's slices. A single 5 GB file can be read by only one slice, leaving the rest of the cluster idle and making the load effectively serial. Splitting the source into many files, ideally a multiple of the total slice count, allows all slices to parse and insert concurrently and is the standard remedy for slow Redshift loads.

Answer

The CSV file uses a delimiter that does not match the table's column definitions.

Answer

The target table's distribution key does not match the sort key, preventing parallel writes.

Answer

The COPY command is missing the PARALLEL ON option, which defaults to OFF.

Question 14

A data engineer is performing EDA on a 300 GB Parquet dataset in Amazon S3 and wants to run interactive SQL from a SageMaker Studio notebook, iterating quickly on queries while other teams also query the same data. The engineer wants results returned to the notebook as a DataFrame and wants to avoid moving or copying the data. Which solution best fits?

Accepted Answer

Use the Athena JDBC driver from the notebook to submit SQL, then read the result set into a pandas DataFrame.. Athena reads Parquet directly from S3 with no data movement and exposes a JDBC endpoint that notebooks can query, returning paginated results that load cleanly into pandas. Columnar Parquet plus partition pruning limits scanned bytes, which keeps interactive queries fast and cheap, and because the service is serverless and shared, multiple teams can query the same catalog tables concurrently without provisioning anything.

Answer

Start an Amazon EMR cluster, run Spark SQL, and write results back to S3 for the notebook to read.

Answer

Create an Amazon Redshift cluster, load the Parquet files with COPY, and connect from the notebook.

Answer

Run aws s3 sync to copy the Parquet files into the notebook's Amazon EBS volume, then query with pandas.

Question 15

A data scientist examines a 20-feature dataset and finds that two features have a Pearson correlation of 0.98, but a scatter plot shows a clear curved (nonlinear) relationship and the features are heavily right-skewed. The scientist wants a single numeric summary that captures monotonic association and is robust to the skew. Which statistic should be reported?

Accepted Answer

Spearman's rank correlation coefficient between the two features.. Spearman's rank correlation is the appropriate summary because it ranks the data before correlating, making it robust to right skew and capable of detecting any monotonic, including nonlinear, association. It yields one interpretable number between -1 and 1 that aligns with the curved scatter plot the scientist observed.

Answer

The Pearson correlation coefficient computed on the raw skewed values.

Answer

The covariance between the two features after mean-centering only.

Answer

The coefficient of determination (R-squared) from a simple linear regression of one feature on the other.

Question 16

A machine learning engineer is training a linear regression model on a dataset with 20 features using Amazon SageMaker's built-in Linear Learner algorithm. The training job completes quickly, but during validation, the model's predictions are consistently poor, with high mean squared error. The engineer suspects that the features have vastly different scales and distributions. Which preprocessing step should be applied to the training data to most likely improve the model's performance?

Accepted Answer

Standardize the features by subtracting the mean and dividing by the standard deviation.. The scenario describes a linear model struggling with features on different scales, a common problem that leads to slow convergence and poor performance. Standardizing features to zero mean and unit variance ensures that all features contribute equally to the model's learning process. This preprocessing step is a standard recommendation for linear models and directly addresses the suspected cause of high error.

Answer

Apply one-hot encoding to all numerical features.

Answer

Increase the number of epochs for training.

Answer

Remove outliers from the dataset using a z-score threshold.

Question 17

A machine learning engineer needs to deploy a model on Amazon SageMaker for real-time inference. The model requires a custom inference script that preprocesses input data. The engineer wants to ensure that the endpoint can automatically scale based on traffic and that the model can be updated without downtime. Which deployment approach should the engineer use?

Accepted Answer

Deploy the model to a SageMaker endpoint with an inference pipeline that includes a preprocessing container and the model container, and configure automatic scaling.. SageMaker inference pipelines enable deploying multiple containers, such as preprocessing and model containers, behind a single endpoint. This supports custom preprocessing while providing real-time inference. The endpoint can be configured with automatic scaling to handle traffic fluctuations, and model updates can be performed with minimal downtime using endpoint updates. Other options either do not provide real-time inference, add unnecessary complexity, or deviate from SageMaker's managed capabilities.

Answer

Deploy the model to an AWS Lambda function with a SageMaker endpoint as a backend, and use Lambda provisioned concurrency for scaling.

Answer

Deploy the model to a SageMaker batch transform job and invoke it on a schedule to simulate real-time inference.

Answer

Deploy the model to an Amazon ECS cluster with a custom application that handles preprocessing and inference, and use ECS service autoscaling.

Question 18

A data engineer needs to catalog metadata for a large collection of Parquet files stored in Amazon S3 so that multiple analytics services can discover and query the tables consistently. The engineer wants automatic schema discovery and a shared, Hive-compatible metastore without running any servers. Which service should be used?

Accepted Answer

AWS Glue Data Catalog. The AWS Glue Data Catalog is the central, serverless, Hive-compatible metadata repository for the AWS analytics ecosystem. Crawlers populate it automatically, and services including Athena, Redshift Spectrum, EMR, and Lake Formation reference the same table definitions. Workgroups, Lake Formation permissions, and S3 Inventory serve different purposes and cannot replace the shared catalog for schema discovery and cross-service querying.

Answer

Amazon S3 Inventory reports

Answer

AWS Lake Formation permissions

Answer

Amazon Athena workgroups

Question 19

A data scientist is using Amazon SageMaker to train a model and needs to monitor the training job's resource utilization and detect anomalies such as high CPU usage or memory leaks. The data scientist wants to receive alerts when the training job's CPU utilization exceeds 90% for more than 5 minutes. Which AWS service should be used to set up these alerts?

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

Accepted Answer

RecordIO. The SageMaker built-in Image Classification algorithm is designed to work with RecordIO (or image files with a list file). RecordIO is the preferred format because it stores multiple images in a single file, improving I/O efficiency during training. Other formats like TFRecord, CSV, and Parquet are not natively supported for this algorithm.

Answer

CSV

Answer

Parquet

Answer

TFRecord

Question 20

A financial services company is training a fraud detection model using Amazon SageMaker. The dataset contains 10 million transactions, with only 0.2% labeled as fraudulent. The team wants to optimize the model's ability to identify fraud while minimizing false positives. Which SageMaker built-in algorithm and evaluation metric should they use to achieve this goal?

Accepted Answer

Use the XGBoost algorithm with the area under the precision-recall curve (AUPRC) as the evaluation metric.. For highly imbalanced fraud detection, XGBoost combined with AUPRC is optimal. XGBoost can model complex non-linear relationships and supports class weighting to handle imbalance. AUPRC focuses on the precision-recall trade-off for the positive class, providing a more informative evaluation than ROC AUC or accuracy when the positive class is rare. This directly aligns with the need to catch fraud while controlling false positives.

Answer

Use the Linear Learner algorithm with the F1 score as the evaluation metric.

Answer

Use the K-Nearest Neighbors algorithm with accuracy as the evaluation metric.

Answer

Use the XGBoost algorithm with the area under the ROC curve (AUC) as the evaluation metric.