AWS Certified Machine Learning Specialty Exam Questions 2021

AWS Certified Machine Learning Specialty Sample Exam Questions

/15

Sample Exam (15 Questions): AWS Machine Learning Specialty

This is free Sample Practice Test. You will have 30 mins to complete this test. Please purchase to get lifetime access to all 204 Real exam questions.

Please fill your Name and Email address and click on next to Start the Exam

1 / 15

You work as a machine learning specialist for a state highway administration department. Your department is trying to use machine learning to help determine the make and model of cars as they pass a camera on the state highways. You need to build a machine learning model to accomplish this problem.
Which modeling approach best fits your problem?

A. Multi-Class Classification

B. Simulation-based Reinforcement Learning

C. Binary Classification

D. Heuristic Approach

2 / 15

You work as a machine learning specialist at a hedge fund. You are working on a time-series price prediction model for the firm and you have setup a data delivery stream using Amazon Kinesis Data Streams. You are creating the data producer application code to take trade data from your trade system and send the trade records to your Kinesis Data Stream. Your python code is structured as follows:
import boto3
import requests
import json
client = boto3.client(‘kinesis’, region_name=’us-east1’)
while True:
             r = requests.get(‘https://trading-applicatio-url’)
             data = json.dumps(r.json())
             client.put_record(
                         parameters needed for put_record api call
             )
             ...
Which of the following options are valid put_record request parameters? Select 3.

A. Data

B. ImplicitHashKey

C. ExplicitHashKey

D. PartitionKeys

E. SequenceNumberForOrdering

F. ShardId

3 / 15

You work for a mining company where you are responsible for the data science behind identifying the origin of mineral samples. Your data origins are Canada, Mexico, and the US. Your training data set is imbalanced as such:
Canada           | Mexico         | US    |
   1,210            |   120              | 68     |
You run a Random Forest classifier on the training data and get the following results for your test data set (your test data set is balanced):
Confusion matrix:
                                         Predicted                      _
Observed        | Canada | Mexico | US | Accuracy |
Canada           |     45      |        3   | 0   |    94%      |
Mexico             |       5      |      38   | 5   |    79%      |
US                   |     19      |        8   | 21 |    44%      |
In order to address the imbalance in your training data you will need to use a preprocessing step before you create your SageMaker training job. Which technique should you use to address the imbalance?

A. Run your training data through a preprocessing script that uses the SMOTE (Synthetic Minority Over-sampling Technique) approach

B. Run your training data through a Spark pipeline in AWS Glue to one-hot encode the features

C. Run your training data through a preprocessing script that uses the feature-split technique

D. Run your training data through a preprocessing script that uses the min-max normalization technique

4 / 15

You work in the security department of your company’s IT division. Your company has decided to try to use facial recognition to improve security on their campus. You have been asked to design a system that augments your company’s building access security by scanning the faces of people entering their buildings and recognizing the person as either an employee/contractor/consultant, who is in the company’s database, or visitor, who is not in their database.
Across their many campus locations worldwide your company has over 750,000 employees and over 250,000 contractors and consultants. These workers are all registered in their HR database. Each of these workers has an image of their face stored in the HR database. You have decided to use Amazon Rekognition for your facial recognition solution. On occasion, the Rekognition model fails to recognize visitors to the buildings. What could be the source of the problem?

A. Face landmarks filters set to a max sharpness

B. Bounding box and confidence score for face comparison threshold tolerances set to max values

C. Confidence threshold tolerance set to the default

D. Face collection contents

5 / 15

You work for a startup ecommerce site that sells various consumer products. Your company has just launched their ecommerce website. The site provides the capability for your users to rate their purchases and the products they have purchased from your ecommerce site. You would like to use the review data to build a recommender machine learning model.
Since your ecommerce site is very new, you don’t yet have a very large review dataset to use for your recommendation model. You have decided to use the Amazon Customer Reviews dataset available from the AWS website as a first data source for your machine learning model. Since your website sells similar products to the products sold on Amazon, you will use the Amazon Customer Reviews dataset as the basis for your initial training runs of your model. Once you have enough data from your own ecommerce site you’ll use that data.
Your goal is to perform sentiment analysis on the review dataset to create your own dataset that will be the source used for your recommender machine learning model. Which set of AWS services would you use to build your data pipeline to produce your sentiment dataset for use by your SageMaker model?

A. S3 -> AWS Glue ETL -> Comprehend -> S3 -> SageMaker

B. S3 -> AWS Glue ETL -> Comprehend -> S3 -> Athena -> QuickSite -> SageMaker

C. S3 -> Kinesis Data Firehose -> Comprehend -> S3 -> SageMaker

D. S3 -> Kinesis Data Firehose -> Lambda -> S3 -> SageMaker

6 / 15

You work as a machine learning specialist for a company that runs car rating website. Your company wants to build a price prediction model that is more accurate than their current model, which is a linear regression model using the age of the car as the single independent variable in the regression to predict the price. You have decided to add the horse power, fuel type, city mpg (miles per gallon), drive wheels, and number of doors as independent variables in your model. You believe that adding these additional independent variables will give you a more accurate prediction of price.
Which type of algorithm will you now use for your prediction?

A. Logistic Regression

B. Decision Tree

C. Naive Bayes

D. Multivariate Regression

7 / 15

You work for a financial services firm that wishes to further enhance their fraud detection capabilities. The firm has implemented fine grained transaction logging for all transactions their customers make using their credit cards. The fraud prevention department would like to use this data to produce dashboards to give them insight into their customer’s transaction activity and to provide real-time fraud prediction.
You plan to build a fraud detection model using the transaction observation data with Amazon SageMaker. Each transaction observation has a date-time stamp. In its raw form, the date-time stamp is not very useful in your prediction model since it is unique. Can you make use of the date-time stamp in your fraud prediction model, and if so how?

A. No, you cannot use the date-time stamp since this data point will never occur again. Unique features like this will not help identify patterns in your data.

B. Yes, you can use the date-time stamp data point. You can just use feature selection to deselect the date-time stamp data point, thus dropping it from the learning process.

C. Yes, you can use the date-time stamp data point. You can transform the date-time stamp into features for the hour of the day, the day of the week, and the month.

D. No you cannot use the date-time feature since there is no way to transform it into a unique data point.

8 / 15

You work for a car rental firm in their car tracking department. Your team is responsible for building the machine learning solutions to track the company's fleet of cars. Each car is equipped with a GPS vehicle tracking device that emits IoT data. You are building a data transformation solution to take the GPS IoT data and transform it before storing it in S3 for use in your machine learning models.
You have decided to use Kinesis Data Firehose data transformation to pre-process your IoT data before storing it in S3. You have written your lambda function that pre-processes the data and you are now testing your data transformation process flow. When running your tests you see that Kinesis Data Firehose is rejecting every record as a data transformation failure. What could be the reason for the failure?

A. In your lambda function you have set the result to OK or Dropped for each record processed.

B. The transformed records from your lambda function consist of the recordId and result parameters.

C. When creating your lambda function you used a lambda blueprint for data transformation from the AWS Serverless Application Repository.

D. When creating your lambda function you used a lambda blueprint for data transformation from the AWS Lambda console.

9 / 15

You work for a flight diagnostics company that builds instrumentation for airline manufacturers. Your company’s instrumentation hardware and software is used to detect flight pattern information such as flight path deviation, as well as airline component malfunction. Your team of machine learning specialists has created a model using the Random Cut Forest algorithm to be used to identify anomalies in the data. The streaming data that your instrumentation processes needs to be cleaned and transformed via feature engineering before passing it to your inference endpoint. You have created the pre-processing and post-processing steps (for cleaning and feature engineering) in your training process.
How can you implement the cleaning and feature engineering steps in your inference processing in the most efficient manner?

A. Execute the pre-processing in a client application before sending the data to your inference endpoint

B. Bundle and export the training pre-processing steps and deploy them to your inference container

C. Bundle and export the training pre-processing steps and deploy them as part of your Inference Pipeline

D. Bundle and export the training pre-processing steps and deploy them to IoT Core on the data emitting devices.

10 / 15

You have just landed a position as a machine learning specialist at a large financial services firm. Your new team is working on a fraud detection model using the SageMaker built-in linear learner algorithm. You are gathering the data required for your machine learning model. The dataset you intend to produce will contain well over 5,000 objects that need to be labeled. Your team wants to control the costs of cleaning your data. Therefore, the team has decided to use SageMaker Ground Truth active learning to automate the labeling of your data.
The Ground Truth automated labeling job initially follows this set of steps:
Selects a random sample of data
sends the sample data to human workers
uses the human-labeled data as validation data
runs a SageMaker batch transform using the validation set which generates a quality metric used to estimate the potential quality of auto-labeling the rest of the unlabeled data
runs a SageMaker batch transform on the unlabeled data
data where the expected quality of automatically labeling the data is above the requested level of accuracy is labeled
After performing the above steps, what does Ground Truth do next to complete the labeling of ALL of your data?

A. Selects a new sample of unlabeled data and sends it to human workers; it uses the existing labeled data to verify the new human-labeled data; repeats this later set of steps until all the data in the dataset is labeled

B. Selects a new sample of unlabeled data and sends it to human workers; it uses the existing labeled data and the new human-labeled data to train a new model; repeats this later set of steps until all the data in the dataset is labeled

C. Selects a new sample of the most hard to identify unlabeled data and sends it to human workers; it uses the existing labeled data to verify the new human-labeled data; repeats this later set of steps until all the data in the dataset is labeled

D. Selects a new sample of the most hard to identify unlabeled data and sends it to human workers; it uses the existing labeled data and the new human-labeled data to train a new model; repeats this later set of steps until all the data in the dataset is labeled

11 / 15

You work for a marketing firm that wants to analyze Twitter user stream data to find popular subjects among users who buy products produced by the firm’s clients. You need to analyze the streamed text to find important or relevant repeated common words and phrases and correlate this data to client products. You’ll then include these topics in your client product marketing material.
Which of the following text feature engineering techniques is the best solution for this task?

A. Orthogonal Sparse Bigram (OSB)

B. Term Frequency-Inverse Document Frequency (tf-idf)

C. Bag-of-Words

D. N-Gram

12 / 15

You work as a machine learning specialist for a sports analytics company. Your company has been contracted by the Major League Baseball Association to perform real-time analytics on baseball statistics as baseball plays unfold live on national television. Your first assignment is to predict the outcome of situational set plays (such as stolen bases or pitch results) as they are about to unfold. Therefore, your model must deliver its predictions in close to real-time.
You have decided to use a SageMaker built-in algorithm. You have looked at classical forecasting methods like autoregressive integrated moving average (ARIMA) and exponential smoothing (ETS) which use one model for each time series in your data. However, you have many time series over which to train.
Based on your performance requirements and your training requirements, which SageMaker built-in algorithm should you use?

A. Linear Learner

B. Neural Topic Model

C. K-Means

D. Random Cut Forest

E. DeepAR Forecasting

F. XGBoost

Option A is incorrect. From the Amazon SageMaker developer guide titled Linear Learner Algorithm “Linear models are supervised learning algorithms used for solving either classification or regression problems.” But you are trying to solve a one-dimensional time series problem so you can extrapolate baseball play time series into the future.
Option B is incorrect. From the Amazon SageMaker developer guide titled Neural Topic Model (NTM) Algorithm “Amazon SageMaker NTM is an unsupervised learning algorithm that is used to organize a corpus of documents into topics that contain word groupings based on their statistical distribution.” So this algorithm is used for natural language processing, not time series problems.
Option C is incorrect. The k-means algorithm is a clustering algorithm. From the Amazon SageMaker developer guide titled K-Means Algorithm “K-means is an unsupervised learning algorithm. It attempts to find discrete groupings within data, where members of a group are as similar as possible to one another and as different as possible from members of other groups.” You are trying to solve a one-dimensional time series problem so you can extrapolate play time series into the future, not a data clustering problem.
Option D is incorrect. From the Amazon SageMaker developer guide titled Random Cut Forest (RCF) Algorithm “Amazon SageMaker Random Cut Forest (RCF) is an unsupervised algorithm for detecting anomalous data points within a data set.” But you are trying to solve a one-dimensional time series problem so you can extrapolate baseball play time series into the future.
Option E is correct. From the Amazon SageMaker developer guide titled DeepAR Forecasting Algorithm “... you have many similar time series across a set of cross-sectional units. For example, you might have time series groupings for demand for different products, server loads, and requests for webpages. For this type of application, you can benefit from training a single model jointly over all of the time series. DeepAR takes this approach. When your dataset contains hundreds of related time series, DeepAR outperforms the standard ARIMA and ETS methods. You can also use the trained model to generate forecasts for new time series that are similar to the ones it has been trained on.” Also from the same developer guide “The training input for the DeepAR algorithm is one or, preferably, more target time series that have been generated by the same process or similar processes. Based on this input dataset, the algorithm trains a model that learns an approximation of this process/processes and uses it to predict how the target time series evolves.” So the DeepAR algorithm is used for one-dimensional time series problems for complex analysis like baseball play prediction.
Option F is incorrect. The XGBoost algorithm is a gradient boosting algorithm. From the Amazon SageMaker developer guide titled XGBoost Algorithm “gradient boosting is a supervised learning algorithm that attempts to accurately predict a target variable by combining an ensemble of estimates from a set of simpler, weaker models.” You are not trying to predict a target value, you are trying to solve a one-dimensional time series problem.
Reference:
Please see the Amazon SageMaker developer guide titled Use Amazon SageMaker Built-in Algorithms, the AWS Machine Learning Blog titled Now Available in Amazon SageMaker: DeepAR algorithm for more accurate time series forecasting, and the AWS StatCast AI page titled See how AI on AWS gives baseball fans new insights into the game

13 / 15

You work for a financial services company where you have a large Hadoop cluster hosting a data lake in your on premises data center. Your department has loaded your data lake with financial services operational data from your corporate actions, order management, cash management, reconciliations, and trade management systems. Your investment management operations team now wants to use data from the data lake to build financial prediction models. You want to use data from the Hadoop cluster in your machine learning training jobs. Your Hadoop cluster has Hive, Spark, Sqoop, and Flume installed.
How can you most effectively load data from your Hadoop cluster into you SageMaker model for training?

A. Use the distcp utility to copy your dataset from your hadoop platform to the S3 bucket where your SageMaker training job can use it

B. Use the HadoopActivity command with AWS Data Pipeline to move your dataset from your hadoop platform to the S3 bucket where your SageMaker training job can use it

C. Use the SageMaker Spark library using the data frames in your Spark clusters to train your model

D. Use the Sqoop export command to export your dataset from your Hadoop cluster to the S3 bucket where your SageMaker training job can use it

14 / 15

You work as a machine learning specialist for a software company that is developing a movie rating social media site where users can rate movies. You want to use your companies data to predict the ratings distribution of a movie based on the genre of the movie. Your training data contains a genre feature with a set of categories such as documentary, romance, etc. You have sorted your data by the genre feature and then used the Amazon ML sequential split option to split your data into training and test datasets.
When using your test dataset to verify your genre-prediction model you discover that the accuracy rate is very low. What could be the underlying problem?

A. You should have sorted by a different feature before you used the sequential split option

B. You should have split your data categorically by genre

C. You should have split your data sequentially by year

D. You should not have used the sequential split option

15 / 15

You work in the machine learning department of a major retail company. Your team is working on a model to predict the region that will have the highest sales for a given quarter. You have selected your observations from past sales cycles for all regions and split your data into training and evaluation datasets. You are now training your linear learner model in Amazon SageMaker and you are trying to select the model hyperparameters that give your team the best predictions.
You have set the predictor_type hyperparameter to binary_classifier. Which loss function hyperparameter setting is NOT one of your options?

A. auto

B. logistic

C. hinge_loss

D. softmax_loss

Your score is

The AWS Certified Machine Learning Specialty Exam Questions certification is intended for individuals who perform a development or data science role. It validates a candidate’s ability to design, implement, deploy, and maintain machine learning (ML) solutions for given business problems.

Abilities Validated by the Certification

Select and justify the appropriate ML approach for a given business problem
Identify appropriate AWS services to implement ML solutions
Design and implement scalable, cost-optimized, reliable, and secure ML solutions

Recommended Knowledge and Experience

1-2 years of experience developing, architecting, or running ML/deep learning workloads on the AWS Cloud
The ability to express the intuition behind basic ML algorithms
Experience performing basic hyperparameter optimization
Experience with ML and deep learning frameworks
The ability to follow model-training best practices
The ability to follow deployment and operational best practices

9 reviews for AWS Certified Machine Learning Specialty Exam Questions 2021

Rated 5 out of 5

Vandana Bansal – May 30, 2020

The explanations of the questions are clear and nicely written. These practice tests and their explanations helped me a lot to pass Machine learning certification. I would highly recommend to anyone planning to take certification.
Rated 5 out of 5

Kristopher Angelo – June 12, 2020

Passed the machine learning certification yesterday and this really helped me a lot. It is harder than the actual exam. It tests your knowledge on each component and it covers all topics. A good gauge to see if you are ready to take the test. I like how well explained the answers were. Plus, it simulates the actual exam environment. I just wish that some questions are less wordy and be direct to the point. But overall, this is the best practice test on the market.
Rated 5 out of 5

Sumitha Roy – September 6, 2020

Thanks a lot Skillcerpro for the great practice test series. It helped me a lot to boost my confidence and pass my exam in first attempt. Mainly the way the explanation is provided for correct and incorrect answers helped me a lot to clear my doubts about many concepts and finally passed.
Rated 5 out of 5

Harris Kinyanjui – September 13, 2020

Excellent is only word coming in my mind after going through the practice test. It really gives me confidence. Best part is the review.
Rated 5 out of 5

Morgane B – September 19, 2020

Practice exam questions were incredibly helpful and I kept on retaking until I was comfortably getting over 90%. I took my exam over the weekend and a number of questions from these practice tests appeared. Passed exam first time
Rated 5 out of 5

Dang Khoi PHAM – October 3, 2020

I passed my exam in the first attempt with 939/1000. I only took 2 over 4 tests because I did not have enough time. Luckily, I learnt from these 2 tests and got my certificate. I am thinking to use your services for my SA Pro.
Rated 5 out of 5

Dinesh Gopinathan – October 20, 2020

The exam questions are challenging and are on par with the aws machine learning Exam questions. The explanations for the choices of each answer are immensely helpful. Definitely recommended.
Rated 5 out of 5

Viranjan Sanjeewa Simon – November 8, 2020

Very well constructed questions with very well constructed answers on why the correct answer is best AND why each of the other answers is not correct or not best. Also, additional material presented in the answer section for each question which reinforces learnings and directs the candidate to authoritative materials for remedial study.
Rated 5 out of 5

Kaitlyn Otto – January 5, 2021

These questions will really help to clear the exam. I took this exam bunch until I scored 80% in all tests. Please read the reasons for wrongs answers , that will help someone to clear. I cleared machine learning recently. Thanks you!!