Cert Notes/ 출퇴근 학습 노트
로드맵
KOEN
CLF-C02 · FoundationalCloud Practitioner - Foundational
DVA-C02 · AssociateDeveloper - Associate
SAA-C03 · AssociateSolutions Architect - Associate
SOA-C02 · AssociateCloudOps Engineer - Associate
SAP-C02 · ProfessionalSolutions Architect - Professional
DOP-C02 · ProfessionalDevOps Engineer - Professional
SCS-C03 · SpecialtySecurity - Specialty
MLA-C01 · AssociateMachine Learning Engineer - Associate
AIF-C01 · FoundationalAI Practitioner - Foundational
DEA-C01 · AssociateData Engineer - Associate
MLS-C01 · SpecialtyMachine Learning - Specialty
  • Week 1
    • 1.ML Lifecycle (Specialty Perspective)
    • 2.ML용 데이터 저장소: S3·EFS·FSx for Lustre·데이터 포맷
    • 3.데이터 수집: Kinesis·Glue·배치 vs 스트리밍
    • 4.Data Labeling: SageMaker Ground Truth·Active Learning·Label Quality
    • 5.Week 1 Integrated Review: ML Overview & Data Engineering 1
  • Week 2
    • 1.데이터 변환과 ETL: AWS Glue, Spark, 그리고 EMR
    • 2.학습 데이터 파이프라인 자동화: Step Functions와 SageMaker Pipelines
    • 3.Data Augmentation and Synthesis: Addressing Insufficient and Imbalanced Data
    • 4.Data Storage and Access Optimization: Pipe vs File Mode, FSx for Lustre, Distributed Training
    • 5.Week 2 Comprehensive Review: From Transformation to Distributed Training Data Supply
  • Week 3
    • 1.Data Cleaning: Missing Values, Outlier Detection, Duplicates and Errors
    • 2.특성 공학: 스케일링, 인코딩, 비닝
    • 3.Time Series, Text Features, and High-Cardinality Categorical Handling
    • 4.SageMaker 도구: Data Wrangler, Processing Job, Feature Store
    • 5.Week 3 Comprehensive Review: Cleaning and Feature Engineering
  • Week 4
    • 1.차원 축소: PCA, t-SNE, 차원의 저주
    • 2.Feature Selection: Filter, Wrapper, Embedded, Importance, Multicollinearity
    • 3.Data Visualization: Distribution, Correlation, QuickSight, Insights
    • 4.Handling Class Imbalance: Over/Undersampling, SMOTE, Class Weights, Evaluation
    • 5.Week 4 Comprehensive Review: Dimensionality, Feature Selection, Visualization, Imbalance
  • Week 5
    • 1.Statistical Foundations: Distribution, Central Tendency, Dispersion, Transformations, Sample and Population
    • 2.Correlation and Relationships: Correlation Coefficients, Causation vs. Correlation, Multivariate Relationships
    • 3.Data Leakage: Causes, Detection, Prevention; Time Series Leakage; Target Leakage
    • 4.Validation Design: train/validation/test Split, Cross-Validation, Time Series Split, Stratified Sampling
    • 5.Week 5 Comprehensive Review: Statistics and Validation Design
  • Week 6
    • 1.알고리즘 선택: 문제 유형별 매핑
    • 2.SageMaker Builtin 1: XGBoost, Linear Learner, K-Means, KNN
    • 3.SageMaker Builtin 2: Text, Image, Time Series, Recommendation
    • 4.Unsupervised/Anomaly Detection: RCF, PCA, IP Insights, Topic Models (LDA/NTM)
    • 5.Week 6 Comprehensive Review: Algorithm Selection and SageMaker Builtins
  • Week 7
    • 1.신경망 기초: 퍼셉트론에서 역전파까지
    • 2.CNN: Convolutional Neural Networks and Computer Vision
    • 3.RNNs and Sequences: From LSTM to Transformer
    • 4.학습 기법과 전이학습
    • 5.Week 7 Synthesis: Deep Learning Summary
  • Week 8
    • 1.SageMaker 학습 작업: Estimator, 입력 모드, 분산 학습, Spot
    • 2.하이퍼파라미터 튜닝(AMT): 베이지안·랜덤·Hyperband
    • 3.Overfitting/Underfitting: Diagnosis and Regularization/Data Augmentation
    • 4.학습 최적화: 배치 크기·학습률, 그래디언트 문제, Debugger/Profiler
    • 5.Week 8 Review: Training, Tuning, Generalization
  • Week 9
    • 1.분류 평가지표: 정확도·정밀도·재현율·F1과 혼동행렬
    • 2.ROC/AUC와 임계값 조정: 곡선으로 읽는 모델 성능
    • 3.회귀 평가지표: RMSE·MAE·MAPE·R²와 잔차 분석
    • 4.모델 디버깅과 편향: SageMaker Debugger와 Clarify
    • 5.Week 9 종합 복습: 평가와 디버깅
  • Week 10
    • 1.추론 옵션: 실시간 vs 서버리스 vs 비동기 vs 배치 변환
    • 2.Real-time Endpoint Operations: Configuration, Auto Scaling, Multi-Model
    • 3.Inference Optimization: Neo, Elastic Inference, Inferentia, Inference Pipelines
    • 4.Deployment Strategies: A/B Testing, Blue/Green, Canary, Shadow, Rollback
    • 5.Week 10 Review: ML Implementation & Operations 1 — Deployment & Inference
  • Week 11
    • 1.Model Monitoring: SageMaker Model Monitor and Drift Response
    • 2.MLOps: SageMaker Pipelines, Model Registry, CI/CD
    • 3.ML Security: IAM Execution Roles, VPC Isolation, KMS Encryption
    • 4.Operations & Cost: Cost Optimization, Logging/Audit, Disaster Recovery
    • 5.Week 11 Review: Monitoring, MLOps, Security, Operations
  • Week 12
    • 1.도메인 1·2 통합 복습: 데이터 엔지니어링 + 탐색적 데이터 분석
    • 2.도메인 3 통합 복습: 모델링(알고리즘·딥러닝·튜닝·평가)
    • 3.도메인 4 + 전체 종합: ML 구현·운영 복습 + 4도메인 교차
    • 4.Synthesis: 4 Domains + End-to-End Scenarios
    • 5.D-Day 마무리: 시험 구성·시간 배분·요구사항 번역표·함정 총정리
합격 후기
← MLS-C01/Week 2/Day 5
MLS-C01· AssociateWeek 2 · Day 5읽기 약 14분

Day 5 - Week 2 Comprehensive Review: From Transformation to Distributed Training Data Supply

📌 핵심 정리

  • Week 2는 데이터 엔지니어링의 두 번째 축 — 변환 → 자동화 → 증강 → 효율적 공급을 다뤘다.
  • Glue = 서버리스·운영 최소, EMR = 세밀 제어·다중 프레임워크. 판단 축은 운영 부담·워크로드 성격·프레임워크 다양성.
  • Step Functions = 범용·혼합 워크플로, SageMaker Pipelines = ML 전용·lineage. 혼합 구성도 정석이다.
  • 불균형은 데이터 층(SMOTE) + 알고리즘 층(가중치·임계값) + 평가 층(PR-AUC) 3층으로 대응한다.
  • 데이터 공급의 목표는 하나다 — 비싼 GPU가 I/O를 기다리며 놀지 않게 하는 것.

Week 2 흐름 한눈에 보기

여기부터는 Pro 전용입니다

Week 1은 누구나 무료로 볼 수 있어요. Week 2부터의 전체 학습 자료와 모의고사·무제한 복습은 Pro 플랜에서 이용할 수 있습니다.

Pro 알아보기로그인
이전Data Storage and Access Optimization: Pipe vs File Mode, FSx for Lustre, Distributed TrainingWeek 2 · Day 4다음 Data Cleaning: Missing Values, Outlier Detection, Duplicates and ErrorsWeek 3 · Day 1