MLS-C01 Practice Test 47 — 15 Questions

Question 1

A data scientist is using SageMaker to train a model. The training job is failing with a 'ResourceLimitExceeded' error. Which action should be taken to resolve this issue?

Accepted Answer

Request a service limit increase for the instance type.. The 'ResourceLimitExceeded' error in SageMaker indicates that the AWS account has reached the maximum number of allowed resources (e.g., instances, vCPUs, or storage) for a given instance type in the current region. Requesting a service limit increase via the AWS Service Quotas console or API directly resolves this by raising the cap for that specific instance type, allowing the training job to proceed.

Answer

Use a different AWS Region.

Answer

Reduce the training dataset size.

Answer

Switch to a different instance type with lower resource requirements.

Question 2

A data scientist wants to query a dataset stored in Amazon S3 using standard SQL without provisioning any servers. The dataset is in CSV format and is updated daily. Which AWS service should be used?

Accepted Answer

Amazon Athena. Amazon Athena is a serverless interactive query service that allows you to query data directly from Amazon S3 using standard SQL without provisioning any servers. It is ideal for querying CSV datasets that are updated daily because it supports schema-on-read, meaning you can define the table schema at query time without loading or transforming the data beforehand.

Answer

Amazon Redshift

Answer

Amazon RDS

Answer

Amazon DynamoDB

Question 3

Refer to the exhibit. A data scientist runs the AWS CLI command shown and gets the output. The scientist wants to create an Athena table over all log files in the 'logs/2023/' prefix, including files smaller than 1000 bytes. Which approach achieves this?

Accepted Answer

Create the table using LOCATION 's3://my-bucket/logs/2023/' which includes all files under that prefix.. Creating an Athena table with the LOCATION pointing to 's3://my-bucket/logs/2023/' will include all objects under that prefix, regardless of size. The CLI command's --query parameter only filters the output of the list-objects command, but does not impose any restriction on the data or the bucket. Option B is incorrect because a WHERE clause in Athena can only filter rows after the table is defined; it cannot include or exclude files from being read. Option C is incorrect because S3 does not have a built-in size restriction on a bucket; the CLI command is just a client-side query. Option D is incorrect because the CLI command is independent; the table can be created directly without needing to modify the CLI command.

Answer

Create the table and add a WHERE clause to include small files.

Answer

Ask the S3 team to remove the size restriction on the bucket.

Answer

Modify the CLI command to remove the size filter and re-run it before creating the table.

Question 4

A company wants to deploy a machine learning model that provides real-time inference with low latency. The model is a small ensemble of three tree-based models. Which Amazon SageMaker approach is most appropriate?

Accepted Answer

Use a SageMaker real-time endpoint with a single inference container.. A SageMaker real-time endpoint with a single inference container is the most appropriate approach because it provides persistent, low-latency inference by keeping the model loaded in memory and handling requests synchronously. For a small ensemble of three tree-based models, a single container can host all models (e.g., using a custom inference script or a multi-model endpoint) and deliver sub-second response times, meeting the real-time requirement.

Answer

Use a SageMaker batch transform job.

Answer

Use AWS Lambda with the model packaged in a layer.

Answer

Use a SageMaker Serverless Inference endpoint.

Question 5

A data engineer is querying the AWS Glue Data Catalog table shown in the exhibit. The engineer runs an Athena query: SELECT * FROM transactions WHERE year=2023. The query returns results quickly. However, a subsequent query: SELECT * FROM transactions WHERE amount > 100 takes a long time. What is the most likely reason for the performance difference?

Accepted Answer

The first query uses a partition column (year), allowing partition pruning, while the second query does not.. The table is partitioned by year and month. The first query filters on a partition column (year), so Athena prunes partitions and scans only the relevant data. The second query filters on a non-partition column (amount), so Athena scans all partitions, resulting in a longer execution time. Option A is incorrect because compression does not directly affect partition pruning; it reduces storage size but not scan time in this context. Option C is incorrect because the data format (Parquet) could help with columnar pruning, but the key difference here is partition pruning, not file format. Option D is incorrect because using SELECT * does not inherently cause slow performance; the lack of partition pruning is the main issue.

Answer

The data is compressed, and the first query benefits from compression.

Answer

The data is stored in Parquet format, which is optimized for columnar access.

Answer

The second query is not optimized because it uses 'SELECT *'.

Question 6

A data scientist is using Amazon SageMaker to perform exploratory data analysis on a dataset with missing values and outliers. Which TWO actions should the scientist take to understand the data quality? (Choose TWO.)

Accepted Answer

Use histograms to visualize the distribution of each numerical feature. Histograms show the distribution of numerical features, helping to identify skewness and outliers. Option E is correct because summary statistics like df.describe() provide count, mean, min, max, and quartiles, which reveal missing values (via count) and outliers (via min/max). Option A is incorrect because a scatterplot matrix visualizes pairwise relationships but does not directly show missing values or outliers. Option C is incorrect because a confusion matrix is used for evaluating classification model performance, not for data exploration. Option D is incorrect because a correlation matrix shows relationships between features but does not highlight missing values or outliers.

Answer

Build a scatterplot matrix to visualize pairwise relationships

Answer

Plot a confusion matrix to assess class separation

Answer

Create a correlation matrix to identify redundant features

Question 7

A data engineer created a CloudFormation template for a Glue ETL job as shown. The job processes 500 GB of data and takes 90 minutes to complete. However, the job fails after 60 minutes. What is the MOST likely cause?

Accepted Answer

The Timeout property is set to 60 minutes, but the job requires more time.. The Glue ETL job has a `Timeout` property set to 60 minutes, but the job requires 90 minutes to complete. When the timeout is reached, AWS Glue forcibly terminates the job, causing it to fail. This is the most direct cause of the failure after exactly 60 minutes of execution.

Answer

The IAM role does not have sufficient permissions.

Answer

The ScriptLocation S3 bucket is in a different region.

Answer

The MaxRetries property is set to 0, so the job does not retry on failure.

Question 8

A data scientist is using Amazon SageMaker to train a deep learning model using a built-in algorithm. The training job uses an ml.p3.2xlarge instance and takes 10 hours to complete. The scientist wants to reduce training time without changing the algorithm or model architecture. The instance's GPU utilization is consistently at 95%, but CPU utilization is only 20%. The data input pipeline uses SageMaker Pipe mode with the 'TrainingInputMode' set to 'Pipe'. The training dataset is 200 GB in CSV format stored in S3. Which approach is most likely to reduce training time?

Accepted Answer

Use a larger instance type with more GPUs, such as ml.p3.8xlarge. GPU utilization is already at 95%, indicating the GPU is the bottleneck. Switching to a larger instance type like ml.p3.8xlarge provides four times the number of GPUs (4 vs. 1), allowing parallel processing of more data and directly reducing wall-clock training time without altering the algorithm or model architecture. The low CPU utilization (20%) confirms that the data pipeline is not a bottleneck, so I/O optimizations are unlikely to help.

Answer

Switch from Pipe mode to File mode to reduce I/O overhead

Answer

Use Pipe mode with 'S3DataType' as 'AugmentedManifestFile'

Answer

Reduce the batch size to improve GPU utilization

Question 9

A financial services company stores sensitive transaction logs in an Amazon S3 bucket. A data engineer must ensure that all objects written to the bucket are encrypted at rest with a customer-managed AWS KMS key. The engineer also needs to enforce that the key is used for every upload and that the encryption is applied automatically without the uploader specifying it. Which solution meets these requirements with the LEAST operational overhead?

Accepted Answer

Enable S3 default encryption with SSE-KMS using the customer-managed key, and modify the bucket policy to deny s3:PutObject requests that specify a different encryption method or no encryption, while allowing the default encryption to apply when no encryption headers are provided.. The scenario requires automatic encryption with a customer-managed KMS key and enforcement that the key is used, with minimal overhead. Enabling S3 default encryption with SSE-KMS applies the key automatically for uploads that do not specify encryption. A bucket policy that denies uploads attempting to use a different encryption method or no encryption ensures the key is used without requiring uploaders to change their requests. This combination meets the requirements with the least operational effort.

Answer

Enable S3 default encryption with SSE-KMS using the customer-managed key, and require that all uploads include the x-amz-server-side-encryption and x-amz-server-side-encryption-aws-kms-key-id headers by adding a bucket policy that denies requests missing these headers.

Answer

Use S3 server-side encryption with Amazon S3 managed keys (SSE-S3) and configure a bucket policy that requires the use of the customer-managed KMS key by checking the aws:RequestTag condition.

Answer

Configure the S3 bucket with default encryption using SSE-KMS with the customer-managed key, and add a bucket policy that denies s3:PutObject requests that do not include the x-amz-server-side-encryption header with value aws:kms.

Question 10

A data scientist runs a SageMaker notebook and uses pandas to explore a dataset. The dataset contains 500,000 rows and 20 columns, including a 'timestamp' column. After loading the data into a DataFrame, the memory usage is unexpectedly high. What is the most likely cause?

Accepted Answer

The default data types inferred by pandas are unnecessarily large for the actual data ranges.. Pandas infers dtypes from the data it reads, and by default it uses int64 for integers and object for strings, which can be far larger than necessary. For 500,000 rows and 20 columns, this can balloon memory usage dramatically — for example, an int64 column uses 8 bytes per value, while an int8 would use 1. Downcasting dtypes is the standard fix.

Answer

The DataFrame created an index column on the timestamp field, doubling memory usage.

Answer

The DataFrame only loaded a sample of the data, but the sample size was too large.

Answer

The CSV file was compressed, and pandas inflated it in memory.

Question 11

A data scientist is training a random forest classifier on Amazon SageMaker and wants to reduce overfitting. Which TWO actions should the scientist take? (Choose TWO.)

Accepted Answer

Limit the maximum depth of trees. Option D is correct because limiting the maximum depth of trees constrains each individual tree's ability to memorize noise in the training data, which is a standard and effective way to reduce overfitting in a random forest. Option E is correct because increasing the number of trees in a random forest reduces variance through averaging of many decorrelated trees, improving generalization without causing overfitting (more trees never hurt accuracy, they only add computation). Option A is incorrect because increasing the number of features considered per split (max_features) makes trees more correlated and more prone to overfitting, not less. Option B is incorrect because increasing the maximum depth allows trees to grow deeper and fit noise, worsening overfitting. Option C is incorrect because decreasing the number of trees reduces the averaging effect and typically increases variance, which can worsen overfitting rather than reduce it.

Answer

Increase the number of features considered per split

Answer

Increase the maximum depth of trees

Answer

Decrease the number of trees

Question 12

A data scientist trains a convolutional neural network (CNN) for image classification. The training loss decreases steadily, but the validation loss starts increasing after 10 epochs. Which technique should the data scientist use to address this problem?

Accepted Answer

Use early stopping to halt training when validation loss stops decreasing.. Early stopping is the appropriate technique because it halts training when validation loss stops decreasing, preventing overfitting. The scenario describes classic overfitting: training loss decreases while validation loss increases after 10 epochs. Early stopping would stop training at the point of best validation performance, saving time and improving generalization.

Answer

Add more data augmentation to the training set.

Answer

Increase the number of training epochs.

Answer

Add more convolutional layers to increase model capacity.

Answer

Increase the learning rate.

Question 13

A company is designing a data pipeline to analyze customer behavior. The pipeline must handle real-time streaming data and batch data. The data must be stored in a data lake on Amazon S3 and also made available for interactive queries. Which THREE services should be combined to build this pipeline? (Choose THREE.)

Accepted Answer

Amazon Kinesis Data Streams. Amazon Kinesis Data Streams is correct because it is the primary AWS service for ingesting and processing real-time streaming data at scale. It can capture and store streaming data from sources like clickstreams or IoT devices, making it available for downstream consumers such as AWS Glue or Amazon Athena for analysis.

Answer

Amazon Redshift

Answer

Amazon DynamoDB Streams

Question 14

A Lambda function is triggered by S3 events. The event payload shown in the exhibit is received by the Lambda function. The function is supposed to process the CSV file and load it into DynamoDB. However, the function fails because it cannot read the file. What is the MOST likely cause?

Accepted Answer

The Lambda function's IAM role does not have s3:GetObject permission. The Lambda function fails because it 'cannot read the file' — that is an S3 read operation, which requires the s3:GetObject permission on the Lambda execution role. Without s3:GetObject, the function's boto3/aws-sdk call to GetObject returns AccessDenied, which matches the symptom. DynamoDB permissions are irrelevant to reading the file.

Answer

The Lambda function lacks DynamoDB write permissions

Answer

The S3 bucket does not exist

Answer

The S3 event notification is misconfigured

Question 15

A machine learning engineer needs to deploy a model that makes real-time predictions with latency under 100ms. The model is a small ensemble of decision trees. Which AWS service is MOST suitable?

Accepted Answer

Amazon SageMaker endpoint. Amazon SageMaker endpoints are purpose-built for real-time inference, providing low-latency (often single-digit to tens of milliseconds) HTTPS endpoints backed by autoscaling instances. For a small ensemble of decision trees, the model is lightweight enough to meet sub-100ms latency easily on a SageMaker endpoint with appropriate instance sizing. SageMaker also supports multi-model endpoints and serverless inference for cost efficiency.

Answer

Amazon EMR with Spark Streaming

Answer

AWS Glue

Answer

AWS Lambda with custom container