MLS-C01 Practice Exam 59 — 20 Questions

Question 1

Which TWO actions are valid ways to handle missing data in a dataset before training a machine learning model? (Select TWO.)

Accepted Answer

Delete rows with missing values. Option A (Delete rows with missing values) is correct because listwise deletion removes incomplete records so the remaining dataset contains only complete cases, which is a standard and valid preprocessing approach when missingness is limited. Option E (Impute missing values with the mean) is correct because mean imputation replaces missing numeric entries with the feature's average, preserving the row count and producing a fully populated dataset suitable for training. Option B is not appropriate as a general rule because dropping every feature with any missing values can discard highly predictive variables and severely reduce the dataset's information. Option C is not a standard valid method because using the maximum value introduces bias and distorts the feature distribution unless there is a specific domain justification. Option D is invalid because most machine learning algorithms cannot natively handle missing values and will error out or produce incorrect results if they are simply ignored.

Answer

Remove all features that have any missing values

Answer

Replace missing values with the maximum value

Answer

Ignore missing values and train the model

Question 2

During EDA, a data scientist plots the distribution of a feature and sees a bimodal pattern. What does this likely indicate?

Accepted Answer

The data may contain two distinct groups.. A bimodal distribution shows two distinct peaks in the feature's probability density. This typically indicates that the data is a mixture of two subpopulations, each with its own central tendency. For example, a dataset of heights might show separate peaks for males and females. Recognizing this is crucial because it suggests the feature may be influenced by a hidden categorical variable, and modeling the data as a single distribution could lead to poor performance.

Answer

The feature has missing values.

Answer

The feature contains outliers.

Answer

The feature needs to be standardized.

Question 3

A data scientist is exploring a dataset with 500 features and 100,000 observations for a regression problem. The scientist notices that many features are highly correlated with each other. Which technique should the scientist use to reduce multicollinearity and improve model interpretability during exploratory data analysis?

Accepted Answer

Calculate Variance Inflation Factor (VIF) for each feature and remove those with VIF > 10.. Variance Inflation Factor (VIF) directly quantifies multicollinearity by measuring how much a feature's variance is inflated by correlation with other features. A VIF above 10 (some use 5) indicates problematic multicollinearity, so removing those features directly addresses the stated goal of reducing multicollinearity and improving interpretability. This is the textbook diagnostic-and-remove approach for the specific problem described.

Answer

Compute mutual information between each feature and the target, and keep only the top 50 features.

Answer

Apply Principal Component Analysis (PCA) to reduce the feature space.

Answer

Use Lasso regression to select features with non-zero coefficients.

Question 4

A machine learning engineer is deploying a model on Amazon SageMaker. Which TWO steps are required to create a SageMaker endpoint?

Accepted Answer

Create a SageMaker model. To deploy a model on Amazon SageMaker, the engineer must first create a SageMaker model (Option A), which packages the model artifacts and the Docker container image with the inference code, along with the IAM role, into a deployable entity that SageMaker can host. The engineer must then create an endpoint configuration (Option D), which specifies the model to deploy, the instance type and count, and any production variants, before calling CreateEndpoint to launch the actual endpoint. Submitting a training job (Option B) is not required at deployment time because the model artifacts may already exist or come from an external source, and training is a separate lifecycle step. Creating a SageMaker pipeline (Option C) is an orchestration construct for automating ML workflows and is not a prerequisite for a single endpoint. Creating a SageMaker notebook instance (Option E) is only an optional development environment and plays no role in the endpoint creation API sequence.

Answer

Submit a training job

Answer

Create a SageMaker pipeline

Answer

Create a SageMaker notebook instance

Question 5

A data scientist is training a deep learning model on a large dataset using Amazon SageMaker. The training job is taking too long and the scientist wants to reduce the training time by distributing the workload across multiple GPUs. Which SageMaker feature should be used to achieve this?

Accepted Answer

Use SageMaker's distributed training libraries. SageMaker's distributed training libraries provide built-in optimized implementations of data parallelism and model parallelism, enabling efficient distribution of deep learning workloads across multiple GPUs with minimal code changes. This directly addresses the need to reduce training time by leveraging multiple GPUs in a SageMaker training job.

Answer

Use Amazon EMR to distribute the training

Answer

Use SageMaker Automatic Model Tuning

Answer

Use SageMaker Hyperparameter Tuning

Question 6

A company is building a sentiment analysis model for customer reviews. The dataset includes 10,000 positive and 10,000 negative reviews. The data scientist splits the data into 70% training, 15% validation, and 15% test sets. After training, the model achieves 99% accuracy on training set but only 82% on validation set. What is the most likely issue?

Accepted Answer

The model is overfitting. A large gap between training accuracy (99%) and validation accuracy (82%) is the classic signature of overfitting — the model has memorized the training data, including noise, and fails to generalize to unseen data. With 20,000 balanced examples, the dataset is not imbalanced, and the model is clearly not underfitting since training performance is near-perfect. The remedy is regularization, dropout, early stopping, or more training data.

Answer

There is data leakage from validation to training

Answer

The dataset is imbalanced

Answer

The model is underfitting

Question 7

An IAM policy is attached to a SageMaker execution role. A data scientist tries to create a training job using a custom algorithm stored in an ECR repository. The training job fails with an 'AccessDenied' error when pulling the Docker image from ECR. What is the missing permission?

Accepted Answer

ecr:GetDownloadUrlForLayer and ecr:BatchGetImage on the ECR repository. When SageMaker pulls a custom Docker image from ECR during training job creation, the execution role needs permissions to download the image layers. The required actions are ecr:GetDownloadUrlForLayer (to generate pre-signed URLs for each layer) and ecr:BatchGetImage (to retrieve image metadata and layer manifests). Without these, the 'AccessDenied' error occurs because SageMaker cannot authenticate or fetch the container image from the ECR repository.

Answer

ecr:PutImage on the ECR repository

Answer

s3:GetObject on the ECR repository

Answer

sagemaker:CreateTrainingJob on the ECR resource

Question 8

A data scientist is analyzing a dataset with a timestamp column. The goal is to identify seasonality and trends. Which visualization technique is most suitable?

Accepted Answer

Time series line plot of the target variable over time.. A time series line plot is the standard visualization for identifying trends and seasonality over time. Option B (box plot grouped by day of week) can show distributions but may not reveal trends or seasonality clearly. Option C (scatter plot of target vs. timestamp) can show patterns but may be less clear than a line plot for time series. Option D (heatmap of correlations) is for exploring relationships between features, not for time series analysis.

Answer

Box plot of the target variable grouped by day of week.

Answer

Scatter plot of the target variable vs. the timestamp.

Answer

Heatmap of correlation between all features.

Question 9

A data scientist is analyzing a dataset with numerical features and a binary target variable. The data scientist creates a pairplot and notices that one feature has a bimodal distribution when colored by the target class. What does this observation suggest?

Accepted Answer

The feature is likely predictive of the target.. A bimodal distribution of a feature when colored by the target class suggests the feature separates the two classes into distinct modes — each class clusters around a different peak. This indicates the feature carries discriminative signal and is likely predictive of the target. In a pairplot, this appears as two distinct humps in the feature's histogram, one per class.

Answer

The feature is irrelevant and should be removed.

Answer

The feature contains outliers that need to be removed.

Answer

The feature has missing values that need to be imputed.

Question 10

A company wants to use Amazon SageMaker to train a model using data that is updated daily. The training data is stored in an S3 bucket, and the team wants to automate the training process whenever new data arrives. Which AWS service should be used to trigger the SageMaker training job?

Accepted Answer

AWS Lambda triggered by S3 event notifications. AWS Lambda triggered by S3 event notifications is the most direct and serverless way to automate a SageMaker training job when new data arrives in S3. When an object is uploaded to the S3 bucket, it can trigger a Lambda function that calls the SageMaker CreateTrainingJob API. This provides near real-time automation without polling or scheduling.

Answer

Amazon CloudWatch Events

Answer

Amazon Simple Queue Service (SQS)

Answer

AWS Step Functions with a scheduled trigger

Question 11

A company uses Amazon SageMaker to train and deploy machine learning models. The training data is stored in Amazon S3 (Parquet format, 10 TB). The data scientists have been running training jobs using the File mode input, but the jobs are taking too long due to data download time. They want to reduce the training start-up time and overall training time. Which solution is MOST cost-effective and efficient?

Accepted Answer

Configure the SageMaker training job to use Pipe mode, which streams data directly from S3 without downloading to the instance's local storage.. Pipe mode in SageMaker streams training data directly from Amazon S3 to the training algorithm without first downloading it to the instance's local storage. This eliminates the data download step, significantly reducing startup time and overall training time for large datasets like 10 TB. It is the most cost-effective because it avoids the need for larger instances or additional data transfer acceleration services.

Answer

Use S3 Transfer Acceleration to speed up the data transfer from S3 to the training instance.

Answer

Use larger EC2 instances with more vCPUs and memory to speed up the training process.

Answer

Enable Elastic Fabric Adapter (EFA) on the training instances to improve network throughput.

Question 12

A data scientist is training a binary classifier on an imbalanced dataset where the positive class represents only 2% of the data. The model achieves 99% accuracy but only identifies 5% of actual positives. Which metric should the scientist use to evaluate the model's ability to detect the positive class?

Accepted Answer

Recall. Recall (sensitivity) measures the proportion of actual positives correctly identified, which is the key concern here. Accuracy is misleading due to class imbalance.

Answer

Accuracy

Answer

F1-score

Answer

Precision

Question 13

Which THREE steps should be taken to secure a SageMaker notebook instance that accesses sensitive data? (Select THREE.)

Accepted Answer

Enable encryption at rest for the notebook's EBS volume. Option A is correct because enabling encryption at rest for the notebook's EBS volume protects sensitive data stored on the instance's attached storage using AWS KMS, ensuring data cannot be read if the underlying volume is compromised. Option C is correct because placing the notebook instance inside a VPC with no internet access (i.e., disabling direct internet access and using VPC endpoints or a NAT gateway for required traffic) prevents exfiltration of sensitive data and blocks unauthorized inbound/outbound connections. Option E is correct because attaching an IAM role with least privilege permissions ensures the notebook can only access the specific AWS resources and actions it needs, limiting the blast radius if the instance is compromised. Option B is incorrect because granting root access increases the attack surface and violates the principle of least privilege. Option D is incorrect because allowing direct internet access exposes the notebook to data exfiltration and unauthorized downloads, undermining the security of sensitive data.

Answer

Grant root access to the notebook instance for flexibility

Answer

Allow direct internet access from the notebook for downloading packages

Question 14

A data engineer is performing EDA on a dataset containing user activity logs from a mobile app. The dataset has 10 million rows and includes columns: 'user_id', 'event_type', 'timestamp', 'device_type', and 'session_duration'. The engineer uses Amazon Athena to query the data stored in S3 as CSV files. The engineer runs a query to find the average session_duration per device_type, but the query takes over 5 minutes and scans 100 GB of data. The engineer wants to reduce query cost and improve performance for future EDA. The dataset is not partitioned, and the engineer anticipates frequent queries filtering on 'timestamp' and 'device_type'. Which action will most effectively reduce data scanned?

Accepted Answer

Partition the table by date derived from timestamp and convert to Parquet.. The most effective because it combines partitioning by date (derived from timestamp) and converting to Parquet format. Partitioning by date enables partition pruning for queries filtering on 'timestamp', drastically reducing the amount of data scanned. Parquet provides columnar storage and compression, further minimizing I/O and cost. Option C (Parquet without partitioning) still requires full file scans when filters are applied. Option B (random sampling) sacrifices accuracy for speed, which is undesirable for accurate EDA. Option D (partitioning by device_type) helps only for device_type filters, not for the common timestamp filters mentioned in the scenario.

Answer

Use random sampling to query a subset of data.

Answer

Convert the data to Parquet format and use columnar storage.

Answer

Partition the table by device_type.

Question 15

A data engineer needs to schedule an AWS Glue ETL job to run every hour. Which service should be used to trigger the job?

Accepted Answer

Amazon CloudWatch Events (EventBridge). Amazon CloudWatch Events (now part of Amazon EventBridge) is the correct service for scheduling AWS Glue ETL jobs on a recurring basis, such as every hour. It allows you to create a time-based rule using a cron or rate expression that triggers an AWS Glue job directly as a target, without needing additional compute or orchestration logic.

Answer

AWS Lambda

Answer

AWS Step Functions

Answer

AWS Data Pipeline

Question 16

Which THREE considerations are important when designing a data lake on Amazon S3?

Accepted Answer

Setting up S3 Lifecycle policies to transition data to colder storage. Option A is correct because S3 Lifecycle policies let you automatically transition objects to colder storage classes such as S3 Standard-IA, S3 Glacier Instant Retrieval, or S3 Glacier Deep Archive, which is essential for controlling cost as a data lake grows. Option C is correct because partitioning data by date (for example, s3://bucket/year=2024/month=05/day=01/) lets query engines like Amazon Athena, Redshift Spectrum, or AWS Glue prune irrelevant partitions, dramatically reducing data scanned and improving query performance. Option E is correct because encrypting data at rest with AWS KMS (SSE-KMS) provides centralized key management, auditability via CloudTrail, and compliance for sensitive data stored in the lake. Option B is incorrect because Provisioned IOPS is an EBS volume feature; S3 does not offer provisioned IOPS and scales throughput automatically. Option D is incorrect because S3 is inherently replicated across multiple Availability Zones within a Region, so restricting data to a single AZ is neither possible nor desirable for durability.

Answer

Using Provisioned IOPS for S3

Answer

Using a single Availability Zone for data storage

Question 17

A company needs to ingest real-time clickstream data from thousands of web servers into AWS for near-real-time analytics. The data volume varies and can spike during promotions. Which service should be used to capture and buffer the data before processing?

Accepted Answer

Amazon Kinesis Data Streams. Amazon Kinesis Data Streams is the correct choice because it is designed for real-time data ingestion and buffering of large streams of data, such as clickstream events from thousands of web servers. It provides durable, low-latency storage (up to 365 days retention) and supports multiple consumers for near-real-time analytics, making it ideal for handling variable and spiky data volumes during promotions.

Answer

Amazon SQS

Answer

Amazon Kinesis Data Firehose

Answer

Amazon MQ

Question 18

A data engineer is designing a data pipeline that ingests data from a relational database into a data lake on Amazon S3. The data must be incrementally loaded daily. Which TWO AWS services can be used together to achieve this?

Accepted Answer

AWS Glue. AWS Glue [CORRECT] is the right choice because it provides serverless ETL with crawlers that populate the AWS Glue Data Catalog and jobs that can read only new/changed rows (for example via JDBC bookmarks) and write partitioned Parquet files to Amazon S3, which directly supports incremental daily loads into a data lake. AWS Database Migration Service (DMS) [CORRECT] is also correct because it can perform ongoing (CDC) or scheduled full-load-plus-incremental replication from a relational database to Amazon S3, capturing changes so only new or modified records are delivered daily. Together, DMS can handle change data capture from the source database while Glue transforms and catalogs the landed data in S3. Amazon Kinesis Data Streams is a streaming service for real-time event ingestion, not a relational-database incremental extract tool, so it does not fit this daily batch scenario. Amazon Athena is a serverless query engine over S3 and does not perform ingestion or incremental extraction. Amazon Redshift is a data warehouse, not a mechanism for incrementally loading a relational database into an S3 data lake.

Answer

Amazon Kinesis Data Streams

Answer

Amazon Athena

Answer

Amazon Redshift

Question 19

A data engineer is designing a data pipeline that transforms raw JSON files (each 50-200 KB) in Amazon S3 into Parquet format using AWS Glue. The pipeline must minimize data processing costs and handle a high volume of small files (millions per day). The engineer configures a Glue ETL job with Spark, but the job is slow and expensive due to overhead of reading many small files. Which optimization should the engineer implement to reduce cost and improve performance?

Accepted Answer

Use S3 object grouping or batch operations to combine small files before Glue processing.. The primary performance bottleneck with many small files in S3 is the overhead of listing, opening, and reading each file individually in Spark. By grouping or batching small files into larger objects (e.g., using S3 Batch Operations or a pre-processing step), you reduce the number of input splits and task launches, which dramatically lowers the cost and runtime of the Glue ETL job. This directly addresses the root cause of the inefficiency rather than merely scaling resources.

Answer

Increase the worker type to G.2X for more memory per worker.

Answer

Increase the number of DPUs allocated to the Glue job.

Answer

Change the output format from Parquet to CSV to reduce compression overhead.

Question 20

A data engineer needs to transfer 50 TB of historical data from an on-premises Hadoop cluster to Amazon S3. The company has a 100 Mbps internet connection and a tight deadline of two weeks. Which AWS service should the engineer use to transfer the data most efficiently?

Accepted Answer

AWS Snowball Edge. B is correct because transferring 50 TB over a 100 Mbps connection would take approximately 48 days (50 TB * 8 / 100 Mbps / 86400 seconds/day), far exceeding the two-week deadline. AWS Snowball Edge is a physical data transport device that can securely transfer petabytes of data offline, bypassing network bandwidth constraints entirely. For large datasets and tight deadlines, Snowball Edge is the most efficient AWS service.

Answer

AWS Storage Gateway (Volume Gateway)

Answer

Amazon S3 Transfer Acceleration

Answer

AWS DataSync over the internet