MLS-C01 practice questions 101–110

1. Question 101 · 368 in bank

   Select 2

   A healthcare company is using an Amazon SageMaker notebook instance to develop machine learning (ML) models. The company's data scientists will need to be able to access datasets stored in Amazon S3 to train the models. Due to regulatory requirements, access to the data from instances and services used for training must not be transmitted over the internet. Which combination of steps should an ML specialist take to provide this access? (Choose two.)

   • AConfigure the SageMaker notebook instance to be launched with a VPC attached and internet access disabled.
   • BCreate and configure a VPN tunnel between SageMaker and Amazon S3.
   • CCreate and configure an S3 VPC endpoint Attach it to the VPC.
   • DCreate an S3 bucket policy that allows traffic from the VPC and denies traffic from the internet.
   • EDeploy AWS Transit Gateway Attach the S3 bucket and the SageMaker instance to the gateway.

   Show answer

   Answer: AC

   Explanation

   A and C are correct. To disable direct internet access, you can specify a VPC for your notebook instance. By doing so, you prevent SageMaker from providing internet access to your notebook instance. As a result, the notebook instance can't train or host models unless your VPC has an interface endpoint (AWS PrivateLink) or a NAT gateway and your security groups allow outbound connections. https://docs.aws.amazon.com/sagemaker/latest/dg/appendix-notebook-and-internet-access.html https://docs.aws.amazon.com/sagemaker/latest/dg/notebook-interface-endpoint.html D is wrong. Bucket policy cant be used to deny internet access. It can only enforce access from VPC or VPC endpoint

   Source: Adapted from community discussion — blanco750

   References

   • AWS documentation
   • AWS documentation
   • AWS documentation
   • AWS documentation

   Next question

2. Question 102 · 368 in bank

   Select 2

   A machine learning (ML) specialist at a retail company is forecasting sales for one of the company's stores. The ML specialist is using data from the past 10 years. The company has provided a dataset that includes the total amount of money in sales each day for the store. Approximately 5% of the days are missing sales data. The ML specialist builds a simple forecasting model with the dataset and discovers that the model performs poorly. The performance is poor around the time of seasonal events, when the model consistently predicts sales figures that are too low or too high. Which actions should the ML specialist take to try to improve the model's performance? (Choose two.)

   • AAdd information about the store's sales periods to the dataset.
   • BAggregate sales figures from stores in the same proximity.
   • CApply smoothing to correct for seasonal variation.
   • DChange the forecast frequency from daily to weekly.
   • EReplace missing values in the dataset by using linear interpolation.

   Show answer

   Answer: AC

   Explanation

   CE C. Apply smoothing to correct for seasonal variation: Seasonal variation can have a significant impact on sales data. By applying smoothing techniques such as moving averages or exponential smoothing, the ML specialist can reduce the noise and fluctuations caused by seasonal effects, allowing the model to capture the underlying patterns more effectively. E. Replace missing values in the dataset by using linear interpolation: Missing data can introduce biases and affect the accuracy of the model. Linear interpolation is a common technique for filling in missing values by estimating the missing data points based on the available data. By replacing the missing values, the ML specialist ensures that the model has a complete and representative dataset to learn from.

   Source: Adapted from community discussion — rags1482

   References

   • AWS documentation

   Next question

3. Question 103 · 368 in bank

   A newspaper publisher has a table of customer data that consists of several numerical and categorical features, such as age and education history, as well as subscription status. The company wants to build a targeted marketing model for predicting the subscription status based on the table data. Which Amazon SageMaker built-in algorithm should be used to model the targeted marketing?

   • ARandom Cut Forest (RCF)
   • BXGBoost
   • CNeural Topic Model (NTM)
   • DDeepAR forecasting

   Show answer

   Answer: B

   Explanation

   B is correct. IMO A - No, Random cut forest is for anomaly detection B - Yes, exactly was XGBoost is good for. Binary classification based on a variety of input features C - No, NTM is unsupervised. The problem states the table already has subscription status, therefore we need a supervised algorithm D - No, DeepAR is used for time-series data

   Source: Adapted from community discussion — hichemck

   Next question

4. Question 104 · 368 in bank

   A company will use Amazon SageMaker to train and host a machine learning model for a marketing campaign. The data must be encrypted at rest. Most of the data is sensitive customer data. The company wants AWS to maintain the root of trust for the encryption keys and wants key usage to be logged. Which solution will meet these requirements with the LEAST operational overhead?

   • AUse AWS Security Token Service (AWS STS) to create temporary tokens to encrypt the storage volumes for all SageMaker instances and to encrypt the model artifacts and data in Amazon S3.
   • BUse customer managed keys in AWS Key Management Service (AWS KMS) to encrypt the storage volumes for all SageMaker instances and to encrypt the model artifacts and data in Amazon S3.
   • CUse encryption keys stored in AWS CloudHSM to encrypt the storage volumes for all SageMaker instances and to encrypt the model artifacts and data in Amazon S3.
   • DUse SageMaker built-in transient keys to encrypt the storage volumes for all SageMaker instances. Enable default encryption ffnew Amazon Elastic Block Store (Amazon EBS) volumes.

   Show answer

   Answer: B

   Explanation

   A. NO - you don't want temporary access B. YES - best practice C. NO - CloudHSM is overkill vs. KMS D. NO - transient keys are transient

   Source: Adapted from community discussion — loict

   Next question

5. Question 105 · 368 in bank

   A data scientist is working on a model to predict a company's required inventory stock levels. All historical data is stored in .csv files in the company's data lake on Amazon S3. The dataset consists of approximately 500 GB of data The data scientist wants to use SQL to explore the data before training the model. The company wants to minimize costs. Which option meets these requirements with the LEAST operational overhead?

   • ACreate an Amazon EMR cluster. Create external tables in the Apache Hive metastore, referencing the data that is stored in the S3 bucket. Explore the data from the Hive console.
   • BUse AWS Glue to crawl the S3 bucket and create tables in the AWS Glue Data Catalog. Use Amazon Athena to explore the data.
   • CCreate an Amazon Redshift cluster. Use the COPY command to ingest the data from Amazon S3. Explore the data from the Amazon Redshift query editor GUI.
   • DCreate an Amazon Redshift cluster. Create external tables in an external schema, referencing the S3 bucket that contains the data. Explore the data from the Amazon Redshift query editor GUI.

   Show answer

   Answer: B

   Explanation

   The option that meets these requirements with the LEAST operational overhead is option B: Use AWS Glue to crawl the S3 bucket and create tables in the AWS Glue Data Catalog. Use Amazon Athena to explore the data. AWS Glue is a fully managed ETL service that can automatically discover and catalog metadata about data stored in various data stores, including Amazon S3. By using AWS Glue to crawl the S3 bucket, the data scientist can easily create tables in the AWS Glue Data Catalog, without needing to create or manage any infrastructure. Amazon Athena is an interactive query service that allows querying data stored in Amazon S3 using SQL. By using Amazon Athena, the data scientist can easily explore the data using SQL, without needing to set up any infrastructure.

   Source: Adapted from community discussion — oso0348

   Next question

6. Question 106 · 368 in bank

   A geospatial analysis company processes thousands of new satellite images each day to produce vessel detection data for commercial shipping. The company stores the training data in Amazon S3. The training data incrementally increases in size with new images each day. The company has configured an Amazon SageMaker training job to use a single ml.p2.xlarge instance with File input mode to train the built-in Object Detection algorithm. The training process was successful last month but is now failing because of a lack of storage. Aside from the addition of training data, nothing has changed in the model training process. A machine learning (ML) specialist needs to change the training configuration to fix the problem. The solution must optimize performance and must minimize the cost of training. Which solution will meet these requirements?

   • AModify the training configuration to use two ml.p2.xlarge instances.
   • BModify the training configuration to use Pipe input mode.
   • CModify the training configuration to use a single ml.p3.2xlarge instance.
   • DModify the training configuration to use Amazon Elastic File System (Amazon EFS) instead of Amazon S3 to store the input training data.

   Show answer

   Answer: B

   Explanation

   The correct solution would be to modify the training configuration to use Pipe input mode. This will allow the training data to stream directly into the training instance as it is being consumed, rather than first being downloaded from S3 into the instance's local storage. This can help reduce storage requirements and optimize performance, while also minimizing costs. Using more or larger instances may help with processing power, but it will not address the storage issue, and may even increase costs unnecessarily. Using Amazon EFS may also be an option, but it may come with additional costs and operational overhead.

   Source: Adapted from community discussion — oso0348

   References

   • AWS documentation

   Next question

7. Question 107 · 368 in bank

   A company is using Amazon SageMaker to build a machine learning (ML) model to predict customer churn based on customer call transcripts. Audio files from customer calls are located in an on-premises VoIP system that has petabytes of recorded calls. The on-premises infrastructure has high-velocity networking and connects to the company's AWS infrastructure through a VPN connection over a 100 Mbps connection. The company has an algorithm for transcribing customer calls that requires GPUs for inference. The company wants to store these transcriptions in an Amazon S3 bucket in the AWS Cloud for model development. Which solution should an ML specialist use to deliver the transcriptions to the S3 bucket as quickly as possible?

   • AOrder and use an AWS Snowball Edge Compute Optimized device with an NVIDIA Tesla module to run the transcription algorithm. Use AWS DataSync to send the resulting transcriptions to the transcription S3 bucket.
   • BOrder and use an AWS Snowcone device with Amazon EC2 Inf1 instances to run the transcription algorithm. Use AWS DataSync to send the resulting transcriptions to the transcription S3 bucket.
   • COrder and use AWS Outposts to run the transcription algorithm on GPU-based Amazon EC2 instances. Store the resulting transcriptions in the transcription S3 bucket.
   • DUse AWS DataSync to ingest the audio files to Amazon S3. Create an AWS Lambda function to run the transcription algorithm on the audio files when they are uploaded to Amazon S3. Configure the function to write the resulting transcriptions to the transcription S3 bucket.

   Show answer

   Answer: A

   Explanation

   I think the answer is A. B: Snowcone has limit with 8 TB. C: is AWS on-premises solution, but the company wants to store these transcriptions in an Amazon S3 bucket in the AWS Cloud for model development. D: 100 Mbps cannot handle petabytes datasync.

   Source: Adapted from community discussion — dunhill

   Next question

8. Question 108 · 368 in bank

   A company has a podcast platform that has thousands of users. The company has implemented an anomaly detection algorithm to detect low podcast engagement based on a 10-minute running window of user events such as listening, pausing, and exiting the podcast. A machine learning (ML) specialist is designing the data ingestion of these events with the knowledge that the event payload needs some small transformations before inference. How should the ML specialist design the data ingestion to meet these requirements with the LEAST operational overhead?

   • AIngest event data by using a GraphQLAPI in AWS AppSync. Store the data in an Amazon DynamoDB table. Use DynamoDB Streams to call an AWS Lambda function to transform the most recent 10 minutes of data before inference.
   • BIngest event data by using Amazon Kinesis Data Streams. Store the data in Amazon S3 by using Amazon Kinesis Data Firehose. Use AWS Glue to transform the most recent 10 minutes of data before inference.
   • CIngest event data by using Amazon Kinesis Data Streams. Use an Amazon Kinesis Data Analytics for Apache Flink application to transform the most recent 10 minutes of data before inference.
   • DIngest event data by using Amazon Managed Streaming for Apache Kafka (Amazon MSK). Use an AWS Lambda function to transform the most recent 10 minutes of data before inference.

   Show answer

   Answer: C

   Explanation

   b https://aws.amazon.com/blogs/machine-learning/build-a-predictive-maintenance-solution-with-amazon-kinesis-aws-glue-and-amazon-sagemaker/ c https://aws.amazon.com/blogs/architecture/realtime-in-stream-inference-kinesis-sagemaker-flink/ Both can do what is being asked. C seems possible, we transform the incoming data and directly call inference endpoint from flink app in KDA.

   Source: Adapted from community discussion — drcok87

   References

   • AWS documentation
   • AWS documentation
   • AWS documentation

   Next question

9. Question 109 · 368 in bank

   A company wants to predict the classification of documents that are created from an application. New documents are saved to an Amazon S3 bucket every 3 seconds. The company has developed three versions of a machine learning (ML) model within Amazon SageMaker to classify document text. The company wants to deploy these three versions to predict the classification of each document. Which approach will meet these requirements with the LEAST operational overhead?

   • AConfigure an S3 event notification that invokes an AWS Lambda function when new documents are created. Configure the Lambda function to create three SageMaker batch transform jobs, one batch transform job for each model for each document.
   • BDeploy all the models to a single SageMaker endpoint. Treat each model as a production variant. Configure an S3 event notification that invokes an AWS Lambda function when new documents are created. Configure the Lambda function to call each production variant and return the results of each model.
   • CDeploy each model to its own SageMaker endpoint Configure an S3 event notification that invokes an AWS Lambda function when new documents are created. Configure the Lambda function to call each endpoint and return the results of each model.
   • DDeploy each model to its own SageMaker endpoint. Create three AWS Lambda functions. Configure each Lambda function to call a different endpoint and return the results. Configure three S3 event notifications to invoke the Lambda functions when new documents are created.

   Show answer

   Answer: B

   Explanation

   It's B, you can use Invoke a Multi-Model Endpoint, when you call invoke_endpoint you need to provide which model filw to use. response1 = runtime_sagemaker_client.invoke_endpoint( EndpointName = "MAIN_ENDPOINT", TargetModel = "model1.tar.gz", Body = body) response2 = runtime_sagemaker_client.invoke_endpoint( EndpointName = "MAIN_ENDPOINT", TargetModel = "model2.tar.gz", Body = body) response3 = runtime_sagemaker_client.invoke_endpoint( EndpointName = "MAIN_ENDPOINT", TargetModel = "model3.tar.gz", Body = body) Ref: https://docs.aws.amazon.com/sagemaker/latest/dg/invoke-multi-model-endpoint.html

   Source: Adapted from community discussion — ADVIT

   References

   • AWS documentation
   • AWS documentation

   Next question

10. Question 110 · 368 in bank

    Select 2

    A manufacturing company needs to identify returned smartphones that have been damaged by moisture. The company has an automated process that produces 2,000 diagnostic values for each phone. The database contains more than five million phone evaluations. The evaluation process is consistent, and there are no missing values in the data. A machine learning (ML) specialist has trained an Amazon SageMaker linear learner ML model to classify phones as moisture damaged or not moisture damaged by using all available features. The model's F1 score is 0.6. Which changes in model training would MOST likely improve the model's F1 score? (Choose two.)

    • AContinue to use the SageMaker linear learner algorithm. Reduce the number of features with the SageMaker principal component analysis (PCA) algorithm.
    • BContinue to use the SageMaker linear learner algorithm. Reduce the number of features with the scikit-learn multi-dimensional scaling (MDS) algorithm.
    • CContinue to use the SageMaker linear learner algorithm. Set the predictor type to regressor.
    • DUse the SageMaker k-means algorithm with k of less than 1,000 to train the model.
    • EUse the SageMaker k-nearest neighbors (k-NN) algorithm. Set a dimension reduction target of less than 1,000 to train the model.

    Show answer

    Answer: AE

    Explanation

    A. Correct B. Incorrect. MDS is Non-linear dimensionality reduction method. https://towardsdatascience.com/11-dimensionality-reduction-techniques-you-should-know-in-2021-dcb9500d388b C. Incorrect. This is a classification problem instead of Regression. D. Incorrect. K-means is for Clustering(Unsupervised learning). E. Correct.

    Source: Adapted from community discussion — Jerry84

    References

    • AWS documentation

    Next question

Keyboard:J Kmove between questionsA–Dselect an answerRshow or hide the answer← →change page

MLS-C01 practice, question 101. 368 questions available.