ML platform
SageMaker Data Wrangler
Visual data preparation and feature engineering.
Key points
- Provides a visual interface for importing, cleaning, transforming, and analyzing data.
- Supports feature engineering steps that can feed SageMaker training workflows.
- Helps identify data quality issues before model training.
- Can export data preparation flows into SageMaker processing or pipeline workflows.
- Targets the data preparation stage, not model hosting.
When to use it
- Choose Data Wrangler when the work is cleaning, transforming, or preparing data visually.
- Use it to explore distributions, handle missing values, encode features, and create transformations.
- Use it before training when raw data is not ready for ML.
Exam tips
- Data Wrangler prepares data; Feature Store stores and serves reusable features.
- Canvas is no-code model building; Data Wrangler is visual data preparation.
- Poor data preparation can cause poor model performance even with good algorithms.
- Do not choose Data Wrangler for runtime inference monitoring; choose Model Monitor.