Exam AWS Certified Big Data - Specialty All Questions

View all questions & answers for the AWS Certified Big Data - Specialty exam

Exam AWS Certified Big Data - Specialty topic 1 question 50 discussion

Exam question from Amazon's AWS Certified Big Data - Specialty

Question #: 50
Topic #: 1

[All AWS Certified Big Data - Specialty Questions]

An administrator tries to use the Amazon Machine Learning service to classify social media posts that mention the administrators company into posts that require a response and posts that do not. The training dataset of
10,000 posts contains the details of each post including the timestamp, author, and full text of the post. The administrator is missing the target labels that are required for training.
Which Amazon Machine Learning model is the most appropriate for the task?

A. Binary classification model, where the target class is the require-response post
B. Binary classification model, where the two classes are the require-response post and does-not-require- response
C. Multi-class prediction model, with two classes: require-response post and does-not-require-response
D. Regression model where the predicted value is the probability that the post requires a response

Show Suggested Answer

Suggested Answer: A 🗳️

by muhsin at Aug. 28, 2019, 2:43 p.m.

Disclaimers:

- ExamTopics website is not related to, affiliated with, endorsed or authorized by Amazon.
- Trademarks, certification & product names are used for reference only and belong to Amazon.

Comments

Submit Cancel

muhsin

Highly Voted 3 years, 8 months ago

B is correct. binary classification has two classess.

upvoted 9 times

mattyb123

3 years, 7 months ago

Thanks for the correction i have overlooked this one sadly. https://docs.aws.amazon.com/machine-learning/latest/dg/types-of-ml-models.html Binary Classification Model ML models for binary classification problems predict a binary outcome (one of two possible classes). To train binary classification models, Amazon ML uses the industry-standard learning algorithm known as logistic regression.

upvoted 6 times

balajisush0312

3 years, 7 months ago

Agree with B.Also, the administrator is missing the target labels that are required for training.

upvoted 1 times

...

yuriy_ber

Highly Voted 3 years, 7 months ago

"The administrator is missing the target labels that are required for training.", so you can't train binary classification model. The only option left is D, here is some Use Case: https://medium.com/analytics-vidhya/a-guide-to-machine-learning-in-r-for-beginners-part-5-4c00f2366b90

upvoted 9 times

ME2000

3 years, 7 months ago

Exactly, The answer is D. https://stats.stackexchange.com/questions/55357/binary-classifier-with-training-data-for-one-label-only

upvoted 4 times

ME2000

3 years, 7 months ago

https://hbr.org/2015/11/a-refresher-on-regression-analysis

upvoted 2 times

iamsajal

3 years, 7 months ago

Binary Classification uses Logistic Regression under the hood. So the answer is B.

upvoted 1 times

...

yogesh88

Most Recent 3 years, 6 months ago

Just attempted exam - Option A was Unary classification model, where the target class is the require-response post. I selected B .

upvoted 1 times

...

freedomeox

3 years, 6 months ago

find the exact sceniro in AWS official sample questions but with a different question: Which two options will create valid target label data? A) Ask the social media handling team to review each post and provide the label. B) Use the sentiment analysis NLP library to determine whether a post requires a response. C) Use the Amazon Mechanical Turk web service to publish Human Intelligence Tasks that ask Turk workers to label the posts. D) Using the a priori probability distribution of the two classes, use Monte-Carlo simulation to generate the labels. which makes more sense then this one...

upvoted 2 times

...

freedomeox

3 years, 7 months ago

The key point here is missing the target labels. There are two kinds of machine learning: supervised/unsupervised. if the target labels are missing, then we can't perform supervised learning, because we don't know the right answer. Therefore, A, D are wrong: we don't know the target class or the prediction value of the training, so we can't train. For B, binary classification is one kind of supervised training, which you must provide the ground truth answer of whether the post is response-needed or not. Therefore we left with C. We use unsupervised learning to tell the model, here we have two classes in all these posts, classify them for me.

upvoted 1 times

...

agm84

3 years, 7 months ago

Amazon ML only supports supervised models, meaning it must have a sample of data with the target value info otherwise the model can not be used for training. Not having the target value invalidates all options in this question. That is why i think its A.

upvoted 1 times

...

san2020

3 years, 7 months ago

my selection D

upvoted 2 times

Corram

3 years, 7 months ago

Regression is not suited for classification, you'd at the minimum still transform the predicted probabilities to 0 or 1 - B is better.

upvoted 2 times

...