Prepare data and model features for ML with ease, speed, and accuracy
summary
SageMaker Data Wrangler shows a point‑and‑click pipeline that imports from S3, Athena, Redshift, Snowflake and Delta Lake, previews CSV/Parquet/JSON, applies 300+ built‑in transforms (including SMOTE and time‑series resampling) or custom Pandas/PySpark/SQL code, and runs a hosted job to write to S3. It can run a Quick Model (random forest) to evaluate feature utility. Feature Store provides online (DynamoDB) and offline (S3) stores with time travel, batch getRecord up to 100 records across 10 feature groups, and single‑digit millisecond online reads. Ground Truth offers 30+ labeling workflows.
AI-written summary · about
chapters (16)
- 0:00Intro
- 2:29Common challenges with data preparation
- 4:06Amazon SageMaker overview
- 4:53Amazon SageMaker Data Wrangler
- 7:44Understand data: Advanced visualization
- 12:14Newly launched: Time-series transformers
- 14:14Quickly estimate model accuracy with
- 16:05Easily visualize the steps of your
- 18:39Deploy data preparation workflows
- 34:24Without a feature store
- 36:00With SageMaker Feature Store
- 39:07Feature consistency with online and offline store
- 40:43Feature engineering options
- 42:12Query from offline store for training or batch transform datasets
- 42:50Inference pattern 1: Endpoint-based
- 43:55Inference pattern 2: Application-based
related talks
33mAccelerate data preparation with Amazon SageMaker Data Wrangler
34mPrepare data for machine learning with ease, speed, and accuracy
44mExploratory data analysis and automated feature engineering
30mAmazon SageMaker Feature Store: Store, discover, & share features for ML apps
33mBuild, train, and deploy ML models using Amazon SageMaker
21mAWS On Air ft. Amazon SageMaker Canvas
32mBuild quality ML models easily & quickly with Amazon SageMaker Autopilot
31mUse AutoML to create high-quality models with just a few clicks