• Contents:

Data Wrangler

• 🔧 Data Quality tool
  • Single interface for data exploration, selection, visualization, cleansing, and processing
  • Data preparation, data transformation and feature engineering
• Import Data
• Preview Data
• Visualize Data
• Transform Data

Additional Features

• SQL support
• Quick Model → can quickly spin up a simple model to see if the data makes sense
• Export Data Flow → share data flow with others, potentially in other tools

Feature Store

• 💡 Reminder: training dataset is originally just raw data. But with feature engineering, raw data becomes features, i.e. inputs to ML models used during training and for inference
• 🔧 Centralized place for storing features and their metadata
  • Ingests features from a variety of sources
• Features (no pun intended)
  • Features discoverable within SageMaker Studio
  • Ability to define the transformation of data into feature
  • Data Wrangler can publish features directly into Feature Store
• Screenshot