AWS Certified Machine Learning - Specialty (MLS-C01) Domain 2

Exploratory Data Analysis

Official Exam Guide: Domain 2: Exploratory Data Analysis

Skill Builder: AWS Certified Machine Learning - Specialty Exam Prep

---

Domain Overview

Domain 2 (24%) focuses on sanitizing and preparing data, performing feature engineering, and analyzing and visualizing data for ML.

---

Task 2.1: Sanitize and prepare data for modeling

Key Concepts:

• Identify and handle missing data, corrupt data, stop words
• Format, normalize, augment, and scale data
• Determine if there’s sufficient labeled data
• Use data labeling tools (Amazon Mechanical Turk)

Essential Documentation:

• Amazon SageMaker Data Wrangler
• Amazon SageMaker Ground Truth
• Amazon Mechanical Turk

---

Task 2.2: Perform feature engineering

Key Concepts:

• Extract features from text, speech, images, public datasets
• Feature engineering concepts: binning, tokenization, outliers, synthetic features, one-hot encoding, dimensionality reduction

Essential Documentation:

• SageMaker Data Wrangler Transforms
• SageMaker Processing Jobs
• Amazon Comprehend Developer Guide
• Amazon Rekognition Developer Guide

---

Task 2.3: Analyze and visualize data for ML

Key Concepts:

• Create graphs (scatter plots, time series, histograms, box plots)
• Interpret descriptive statistics (correlation, summary statistics, p-value)
• Perform cluster analysis (hierarchical, elbow plot, cluster size)

Essential Documentation:

• SageMaker Clarify for Data Bias
• Amazon QuickSight User Guide
• Amazon Athena User Guide

---

AWS Service FAQs

• SageMaker Data Wrangler FAQs
• SageMaker Ground Truth FAQs
• Amazon QuickSight FAQs

---

Study Tips

1. Master data cleaning - Handle missing values (imputation, deletion), remove outliers, handle duplicates, address class imbalance (SMOTE, undersampling).

2. Learn feature engineering - Numeric: normalization, standardization, binning. Text: tokenization, TF-IDF, word embeddings. Categorical: one-hot encoding, label encoding.

3. Understand dimensionality reduction - PCA for linear reduction, t-SNE for visualization, feature selection methods (filter, wrapper, embedded).

4. Practice data visualization - Histograms for distributions, scatter plots for correlations, box plots for outliers, heatmaps for correlation matrices.

5. Study descriptive statistics - Mean, median, mode, standard deviation, correlation coefficients, p-values for hypothesis testing.

---

Note: This is Domain 2 of 4, representing 24% of exam content.