book

AWS Certified Machine Learning Engineer Study Guide

June 2025

Intermediate to advanced

448 pages

13h 10m

English

Read now

Unlock full access

Understanding Artificial IntelligenceUnderstanding Machine LearningUnderstanding Deep LearningSummaryExam EssentialsReview Questions

Introducing Ingestion and StorageIngesting and Storing DataSummaryExam EssentialsReview Questions
IntroductionUnderstanding Feature EngineeringData Cleaning and TransformationFeature Engineering TechniquesData LabelingManaging Class ImbalanceData SplittingSummaryExam EssentialsReview Questions
Understanding AWS AI ServicesDeveloping Models with Amazon SageMaker Built-in AlgorithmsCriteria for Model SelectionSummaryExam EssentialsReview Questions
TrainingHyperparameter TuningModel Performance EvaluationDeep-Dive Model Tuning ExampleSummaryExam EssentialsReview Questions
AWS Model Deployment ServicesAdvanced Model Deployment TechniquesOrchestrating ML WorkflowsDeep Dive Model Deployment ExampleSummaryExam EssentialsReview Questions
Monitoring Model InferenceMonitoring Infrastructure and CostSummaryExam EssentialsReview Questions
Security Design PrinciplesSecuring AWS ServicesSummaryExam EssentialsReview Questions
Chapter 1: Introduction to Machine LearningChapter 2: Data Ingestion and StorageChapter 3: Data Transformation and Feature EngineeringChapter 4: Model SelectionChapter 5: Model Training and EvaluationChapter 6: Model Deployment and OrchestrationChapter 7: Model Monitoring and Cost OptimizationChapter 8: Model Security
Linear AlgebraStatisticsProbability TheoryCalculus

Appendix AAnswers to the Review Questions

D. PDF format is generally considered an example of unstructured data rather than semi-structured data. This is because although a PDF may contain text, its structure is not easily parsed by machines, making it difficult to extract meaningful data without specialized tools. NoSQL, JSON, and XML are all examples of semi-structured data because they have a flexible schema and can contain both structured and unstructured information. Therefore, answers A, B, C are incorrect.
D. A model must always be trained before evaluating its performance. During training, the model learns patterns and relationships within the data, adjusting its parameters to minimize ...