Purpose of Quiz Walkthroughs:
- Exposure to Exam Questions: Get familiar with the type and difficulty of questions you might encounter on the actual exam.
- Helpful Methods: Learn strategies for handling multiple-choice questions, focusing on key elements, and spotting incorrect answers.
- Identify Knowledge Gaps: Reinforce confident areas and highlight areas to focus on in your studies.
Approach to Each Question:
- Read the Question: Carefully review the entire question.
- Identify Keywords: Focus on key phrases that guide you toward the answer.
- Evaluate Options: Go through each answer choice, select the best answer, and discuss why it’s correct or incorrect.
Question 1:
Question: A company wants to analyze diverse data sources (audio, text feedback, images, scanned documents) to improve business operations. Historic audio files from call centers stored in Amazon S3.
The company wants to identify historic audio files with its brand and convert them to text for further analysis. The files are stored in Amazon S3. Which combination of steps will meet these requirements? Select two.
Keywords:
- Identify audio files, convert to text, stored in Amazon S3.
Options:
- A: Create a batch transcription job in Amazon Transcribe.
- B: Create a real-time transcription job in Amazon Transcribe.
- C: Create a transcription job in Amazon Translate.
- D: Use custom vocabularies to improve transcription accuracy.
- E: Use custom language models to improve transcription accuracy.
Evaluation of Options:
- Option A (Correct):
Amazon Transcribe can convert speech into text. Since the audio files are historic and stored in S3, use a batch transcription job. - Option B (Incorrect):
Real-time transcription is for live data streams, not for historic files stored in S3. - Option C (Incorrect):
Amazon Translate is for translation, not speech-to-text conversion. - Option D (Correct):
Custom vocabularies in Amazon Transcribe improve accuracy for domain-specific terms like brand names. This is ideal for identifying the company’s brand. - Option E (Incorrect):
A custom language model is used for complex or large vocabularies, which isn’t needed in this case for brand identification.
Final Answer: A and D
- These are the correct steps to meet the requirements of converting audio files to text and improving accuracy.
Question 2 :
Question: A company wants to gain insights from diverse data sources and formats to improve business operations, including:
Audio from call centers
Text feedback from customers
Product images
Scanned documents
Which AWS service can identify the global sentiment of customer feedback from text?
Keywords:
- Identify, global sentiment, text, customer feedback.
Options:
- A: Amazon Translate
- B: Amazon Comprehend
- C: Amazon Transcribe
- D: Amazon Polly
Evaluation of Options:
- Option A (Incorrect):
Amazon Translate is used for translation between languages, not for identifying sentiment. - Option B (Correct):
Amazon Comprehend is a natural language processing (NLP) service that can analyze text and identify sentiment, making it the right choice for this question. - Option C (Incorrect):
Amazon Transcribe is used to convert speech into text. It cannot identify sentiment. - Option D (Incorrect):
Amazon Polly is a text-to-speech service. It does not analyze sentiment in text.
Final Answer: B (Amazon Comprehend)
- Amazon Comprehend is the correct service to identify sentiment in customer feedback text.
Question 3:
Question: A data scientist is building an ML model. Select the correct ML development stage for each step.
The stages: Feature Engineering, Model Deployment, Model Evaluation, and Model Monitoring. Match the ML development stages to the given steps.
Prompts to Match:
- Perform explainability techniques to identify the accuracy of model-generated results.
- Identify data quality issues, model quality issues, bias drift, or feature attribution drift.
- Release the model into production so the model can begin making predictions.
- Select and transform variables to enhance a training dataset.
Evaluation of the Prompts:
Perform explainability techniques to identify the accuracy of model-generated results
- Match: Model Evaluation
- Explanation: After model training, you evaluate accuracy and performance during this stage. The goal is to assess if fine-tuning is needed or if the model is ready for deployment.
Identify data quality issues, model quality issues, bias drift, or feature attribution drift
- Match: Model Monitoring
- Explanation: After deployment, this stage involves monitoring the model for data and model quality issues. It checks for performance degradation or drift in bias and features.
Release the model into production so the model can begin making predictions
- Match: Model Deployment
- Explanation: After training and evaluation, the model is deployed into production to start making predictions.
Select and transform variables to enhance a training dataset
- Match: Feature Engineering
- Explanation: In the data preparation stage, variables are selected and transformed to create features that enhance model training and improve accuracy.
Next step is to reflect on any knowledge gaps identified during these questions.
