MLS-C01 Practice Exam 52 — 20 Questions

Question 1

A machine learning team is using SageMaker to train a model with a custom Docker container. The training script runs locally but fails on SageMaker with a 'Permission denied' error when writing to /opt/ml/model. What is the likely cause?

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

Accepted Answer

The container's user does not have write permission to /opt/ml/model. SageMaker mounts the model artifact directory at /opt/ml/model inside the training container and expects the training script to write model artifacts there. If the Docker container runs as a non-root user (e.g., USER appuser) and that user lacks write permissions on /opt/ml/model, the script will fail with 'Permission denied'. This is a common issue when custom containers are built with restrictive file ownership, since SageMaker does not automatically chown the mounted directories to the container's user.

Answer

The Docker image is too large

Answer

The training script is trying to read from /opt/ml/input/data instead of /opt/ml/input/data/training

Answer

The training data is not in the correct S3 bucket

Question 2

A company uses SageMaker to run training jobs on a schedule. The training data is stored in an S3 bucket that receives new data every hour. Which TWO approaches can the company use to trigger a training job when new data arrives?

Accepted Answer

Set up an Amazon EventBridge rule that triggers on S3 Object Created events and targets a Lambda function. Option B is correct because Amazon EventBridge natively supports S3 events (Object Created) as an event source, and an EventBridge rule can match those events and invoke a Lambda function as a target, which can then call the SageMaker CreateTrainingJob API to start the training job when new data lands. Option E is correct because S3 event notifications can be configured directly on the bucket to invoke a Lambda function on s3:ObjectCreated:* events, and that Lambda function can start the SageMaker training job, providing a direct event-driven trigger. Option A is not appropriate because SQS buffering plus polling from a training instance is a pull-based pattern that does not itself trigger a job on arrival and adds unnecessary complexity. Option C is wrong because a scheduled Step Functions execution is time-based, not event-driven, so it cannot react to new data arriving. Option D is wrong because CloudWatch Logs monitoring S3 access logs is indirect, delayed, and not a supported real-time trigger mechanism for object creation events.

Answer

Use an SQS queue to buffer S3 events and poll from a training instance

Answer

Use AWS Step Functions with a scheduled execution

Answer

Use CloudWatch Logs to monitor S3 access logs and trigger a Lambda function

Question 3

During EDA, a data scientist notices that a numeric feature 'age' has outliers beyond 3 standard deviations. What is the most appropriate first step?

Accepted Answer

Investigate the source of the outliers. The most appropriate first step when encountering outliers beyond 3 standard deviations is to investigate their source before deciding on any treatment. Outliers may represent data entry errors, measurement issues, or genuinely rare but valid events that carry important signal. Understanding their origin informs whether to remove, transform, or retain them, aligning with best practices in exploratory data analysis (EDA).

Answer

Use the feature as-is in the model

Answer

Apply a log transformation to the feature

Answer

Remove all rows with outlier values

Question 4

A data scientist is training a binary classification model on an imbalanced dataset (95% negative class, 5% positive class). The model currently achieves 94% accuracy but a recall of only 0.10 on the positive class. Which TWO strategies should the data scientist consider to improve recall without significantly sacrificing precision? (Choose 2.)

Accepted Answer

Assign higher class weights to the positive class in the loss function.. Option C is correct because assigning higher class weights to the positive class in the loss function makes errors on the minority (positive) class more costly, directly pushing the model to increase recall on that class while still balancing precision through the weighted objective. Option E is correct because SMOTE (Synthetic Minority Over-sampling Technique) generates synthetic positive-class examples via k-nearest-neighbor interpolation, enlarging the minority class and giving the model more positive signal to learn from, which typically raises positive-class recall without the severe precision loss of naive duplication. Option A is not ideal here because undersampling the majority class to 5% discards ~90% of the negative data, which can hurt precision and overall generalization. Option B is unrelated: increasing regularization strength reduces overfitting but does not address class imbalance and can even lower recall by shrinking the model's fit. Option D is also unrelated: adding layers increases capacity but does not correct the class imbalance and may worsen overfitting on the minority class.

Answer

Undersample the majority class to match the minority class size.

Answer

Increase the regularization strength to reduce overfitting.

Answer

Use a deeper neural network with more layers.

Question 5

A data analyst is using Amazon QuickSight to explore a dataset with 10 million rows. The analyst wants to create a histogram of a numerical column. However, the query is taking too long. Which action should the analyst take to improve performance without losing accuracy?

Accepted Answer

Import the dataset into SPICE (Super-fast, Parallel, In-memory Calculation Engine).. Importing the dataset into SPICE is correct because SPICE is QuickSight's in-memory, columnar calculation engine that pre-aggregates and caches data, allowing histogram generation over 10 million rows without repeatedly querying the source. It preserves full accuracy because all rows are loaded, unlike sampling. This is the standard QuickSight performance optimization for large datasets.

Answer

Change the data source to Amazon Athena directly with a limit clause.

Answer

Reduce the number of bins in the histogram.

Answer

Use a sample of the data (e.g., 1 million rows) for the histogram.

Question 6

A company is using SageMaker to host a model for real-time inference. They notice that the endpoint's latency increases over time. The model is stateless and the inference code does not log any errors. What is the MOST likely cause?

Accepted Answer

Memory leak in the inference container. A memory leak in the inference container causes the process's resident memory to grow over time as allocated memory is not freed. Since the model is stateless and no errors are logged, the leak is likely in the inference code or a dependency (e.g., a TensorFlow session or a Python list that grows unbounded). As memory pressure increases, the operating system may swap or the container may be OOM-killed, leading to increased garbage collection pauses and higher latency for each request.

Answer

Gradual increase in request payload size

Answer

Endpoint auto scaling is adding new instances

Answer

Model is accumulating state from previous requests

Question 7

A machine learning team stores training data in an Amazon S3 bucket with versioning enabled. A data engineer must ensure that SageMaker training jobs can read the data but cannot delete or overwrite objects, and that accidental deletions can be recovered. The bucket is in the same account as the SageMaker execution role. Which combination of settings should the engineer configure?

Accepted Answer

Grant the SageMaker execution role only s3:GetObject and s3:ListBucket on the bucket, and enable S3 Versioning plus a bucket policy statement denying s3:DeleteObject for that role.. Limiting the SageMaker execution role to read and list actions enforces least privilege so training jobs can access data but cannot modify it. Adding an explicit deny for delete actions removes any accidental permission path, and enabling S3 Versioning preserves prior versions so an overwrite or deletion can be reversed by restoring an earlier version.

Answer

Enable S3 Object Lock in governance mode with a 30-day retention period on all objects and give the role full S3 permissions.

Answer

Attach an IAM policy that denies s3:DeleteObject only on the specific prefix used for training data, and rely on S3 Versioning for overwrite recovery.

Answer

Use an S3 bucket policy that requires MFA delete and give the SageMaker role permissions to use MFA when deleting objects.

Question 8

A data analyst is exploring a dataset and notices that the target variable has a Poisson distribution. Which type of model is most appropriate for this target?

Accepted Answer

Poisson regression. Poisson regression is the correct choice because it is specifically designed for modeling count data where the target variable follows a Poisson distribution, which is characterized by non-negative integer values and a variance equal to the mean. This aligns directly with the data analyst's observation of a Poisson-distributed target, making Poisson regression the most appropriate generalized linear model (GLM) for this scenario.

Answer

Linear regression

Answer

Cox proportional hazards model

Answer

Logistic regression

Question 9

A data scientist is analyzing a dataset with a target variable that is binary (0/1). Which visualization is most appropriate to explore the relationship between a continuous feature and the target?

Accepted Answer

Box plot of the feature grouped by target.. A box plot of the continuous feature grouped by the binary target is the most appropriate visualization because it shows the distribution (median, quartiles, outliers) of the continuous feature separately for each target class. This makes it easy to see whether the feature's distribution differs between class 0 and class 1, which is exactly the relationship a data scientist wants to explore before modeling.

Answer

Scatter plot of the feature vs. the target.

Answer

Bar chart of the feature.

Answer

Histogram of the feature.

Question 10

During EDA, a data scientist generates a pairplot of the dataset and observes that two features have a Pearson correlation coefficient of 0.95. Which TWO conclusions can the scientist draw from this observation? (Choose 2)

Accepted Answer

The two features may be multicollinear. Option A is correct because a Pearson correlation coefficient of 0.95 between two features indicates a very high degree of linear association, which is a classic sign that the two features may be multicollinear, meaning they carry largely redundant information and can destabilize models such as linear regression. Option B is correct because the Pearson correlation coefficient directly measures the strength and direction of a linear relationship, and a value of 0.95 (close to +1) indicates a strong positive linear relationship. Option C is incorrect because a coefficient of 0.95 is positive, so the features move in the same direction, not opposite directions; a negative value near -1 would indicate an inverse relationship. Option D is incorrect because statistical independence would correspond to a correlation near 0, whereas 0.95 shows a strong dependence. Option E is incorrect because correlation does not imply causation; the high coefficient only shows association, not that one feature causes the other.

Answer

The two features move in opposite directions

Answer

The two features are statistically independent

Answer

One feature causes the other

Question 11

A company is using Amazon SageMaker to train a time series forecasting model using the DeepAR algorithm. The training data contains multiple time series. The model is overfitting. Which action is LEAST likely to reduce overfitting?

Accepted Answer

Reduce the number of time series in the training set.. Reducing the number of time series in the training set reduces the diversity of training data, which typically increases overfitting rather than reducing it. DeepAR relies on learning patterns across multiple related time series to generalize well; fewer time series mean less shared statistical strength, making the model more likely to memorize noise in the remaining series.

Answer

Decrease the number of layers in the neural network.

Answer

Increase the dropout rate.

Answer

Decrease the context length.

Question 12

A data scientist is using SageMaker to train a model and wants to track experiments, including hyperparameters and metrics. Which TWO actions should the scientist take to set up experiment tracking? (Choose TWO.)

Accepted Answer

Use the SageMaker Experiments Python SDK to create an experiment and log runs.. The SageMaker Experiments Python SDK provides the primary interface for creating and managing experiments, allowing the data scientist to log runs, hyperparameters, and metrics in a structured way. This SDK directly integrates with SageMaker training jobs and notebook executions to capture experiment metadata.

Answer

Enable SageMaker Model Monitor to track training metrics.

Answer

Configure CloudWatch Logs to store experiment data.

Answer

Enable SageMaker Studio to automatically capture experiments.

Question 13

A company is using AWS Glue to catalog metadata from various data sources. The crawler is configured to run daily. However, the catalog is not reflecting new partitions added to an S3 bucket during the day. What is the MOST likely cause?

Accepted Answer

The crawler is not scheduled frequently enough to capture changes. The crawler is configured to run daily, but new partitions are being added to the S3 bucket throughout the day. Since the crawler only runs once per day, it will not detect and catalog those new partitions until its next scheduled run. To capture changes more frequently, the crawler schedule should be increased or an event-driven trigger (e.g., using Amazon S3 Events and AWS Lambda) should be implemented.

Answer

The S3 bucket has insufficient permissions for the Glue crawler

Answer

The table schema has changed and the crawler does not update it

Answer

The data format is not supported by AWS Glue

Question 14

A machine learning team is using AWS Glue to prepare data for training. They notice that the ETL job takes a long time to process large datasets. Which change is most likely to improve performance?

Accepted Answer

Increase the number of DPUs for the Glue job.. Increasing the number of DPUs (Data Processing Units) for the AWS Glue job allocates more distributed computing resources, which allows the job to process data in parallel across more executors. This directly reduces the runtime for large datasets by improving the parallelism of Spark transformations and actions.

Answer

Decrease the number of workers in the Glue job.

Answer

Disable Spark shuffle operations.

Answer

Reduce the dataset size by sampling.

Question 15

A data scientist trains a gradient boosting model on a large dataset using SageMaker. The training completes successfully, but when deploying the model to a real-time endpoint, inference latency is too high. Which change is MOST likely to reduce latency without significant accuracy loss?

Accepted Answer

Prune the trees by removing nodes with low importance. Pruning trees by removing nodes with low importance reduces the model's complexity, which directly decreases inference latency because fewer decision paths need to be evaluated. In gradient boosting, this can be done with minimal accuracy loss if the removed nodes correspond to splits that contribute little to the overall prediction, as measured by feature importance or gain.

Answer

Use a larger instance type for the endpoint

Answer

Increase the number of trees in the ensemble

Answer

Use SageMaker Batch Transform instead of real-time

Question 16

A company is using Amazon SageMaker Ground Truth to build a training dataset for an image classification model. The company has a large number of unlabeled images stored in Amazon S3. The data science team wants to use a private workforce consisting of internal employees to label the images. The team creates a labeling job with a private workforce. After starting the job, the team notices that the labeling tasks are not being assigned to any workers. The workers have been added to the private workforce and have received their login credentials. What is the MOST likely cause of this issue?

Accepted Answer

The workers have not been added to the work team that is assigned to the labeling job.. In Amazon SageMaker Ground Truth, a private workforce consists of work teams, and workers must be assigned to a work team that is associated with the labeling job. If workers are added to the private workforce but not to the specific work team used by the labeling job, they will not receive tasks. The most likely cause is that the workers are not in the assigned work team.

Answer

The labeling job is configured for a different task type than image classification.

Answer

The S3 bucket containing the images has incorrect permissions, preventing workers from viewing the images.

Answer

The workers do not have the required IAM permissions to access the labeling portal.

Question 17

A data scientist is building a model to predict customer churn. The dataset includes both numerical features (e.g., account age, usage minutes) and categorical features (e.g., region, plan type). The data scientist wants to use a linear classifier. Which feature engineering step is required before training?

Accepted Answer

One-hot encode categorical features. Linear classifiers (e.g., logistic regression, linear SVM) require numerical input and cannot directly process categorical text labels. One-hot encoding converts each categorical feature into binary indicator columns, allowing the linear model to learn separate weights for each category. Without this step, the model would either fail to train or treat categorical strings as ordinal values, which is mathematically invalid for linear decision boundaries.

Answer

Normalize numerical features

Answer

Impute missing values

Answer

Remove outliers

Question 18

A data scientist notices that a linear regression model trained on a dataset has high variance. The model performs well on the training data but poorly on the test data. Which action is most likely to reduce the variance?

Accepted Answer

Apply L2 regularization to the model. High variance indicates the model is overfitting to the training data. L2 regularization (ridge regression) adds a penalty proportional to the square of the magnitude of the coefficients, which shrinks them toward zero. This reduces the model's sensitivity to noise in the training data, thereby lowering variance and improving generalization to the test set.

Answer

Decrease the amount of training data

Answer

Increase the number of gradient descent iterations

Answer

Add more features to the model

Question 19

An ML engineer runs the AWS CLI command above to list files in a training data bucket. The engineer notices that the three CSV files have different sizes but the same number of columns. What is the MOST likely cause of the size variation?

Accepted Answer

The files contain a different number of rows.. When CSV files share the same schema (same number of columns) but differ in byte size, the most likely explanation is that they contain different numbers of rows. File size scales primarily with row count and the length of values in each row, so a file with 10,000 rows will be larger than one with 5,000 rows even with identical columns.

Answer

The files are compressed with different algorithms.

Answer

Some files have duplicate headers.

Answer

The files have different column data types.

Question 20

A data engineer is tasked with building a system to process a continuous stream of IoT sensor data. The data must be processed in near real-time, and the results must be stored in Amazon S3 partitioned by hour. Which AWS service is the most cost-effective and simplest to implement?

Accepted Answer

Amazon Kinesis Data Firehose. Amazon Kinesis Data Firehose is a fully managed service that ingests streaming data and automatically delivers it to destinations like Amazon S3, with built-in support for dynamic partitioning by keys such as hour. It requires no infrastructure management, scales automatically, and you pay only for the data volume processed, making it the most cost-effective and simplest solution for near real-time ingestion with hourly S3 partitioning. Firehose also handles buffering, compression, and encryption, reducing operational overhead.

Answer

Amazon Simple Queue Service (SQS) with AWS Lambda

Answer

Amazon Kinesis Data Streams with Amazon EC2 consumers

Answer

AWS Database Migration Service (DMS) for continuous replication