Prepare data for machine learning with ease, speed, and accuracy
summary
Demonstrates a concrete workflow: use SageMaker Data Wrangler to load ~2 million CSV rows with 26 columns, run visualizations and 300+ built-in plus custom pandas and PySpark transforms, export a flow that spins up a processing job and creates a feature group in SageMaker Feature Store, ingesting about 1.7 million records (online/offline sync) via a pipeline triggered from S3 in ~6 minutes, then join a prebuilt weather feature group and train an XGBoost model on ~34k Boston records.
AI-written summary · about
chapters (9)
- 0:00Introduction to ML data prep
- 1:25SageMaker Data Wrangler overview
- 3:12SageMaker Feature Store overview
- 6:54Demo 1: Data Wrangler transformation
- 14:35Demo 2: Feature ingestion pipeline
- 20:57SageMaker Pipelines deep dive
- 24:20Working with offline Feature Store
- 27:23Demo 3: Training with shared features
- 30:24Summary and resources
related talks
58mPrepare data and model features for ML with ease, speed, and accuracy
33mAccelerate data preparation with Amazon SageMaker Data Wrangler
44mExploratory data analysis and automated feature engineering
29mAutomatically Build Models With Amazon Sagemaker
30mAmazon SageMaker Feature Store: Store, discover, & share features for ML apps
31mProductizing Machine Learning over Big Data with AWS tools
29mImplementing MLOps practices with Amazon SageMaker
01h 18mFrom raw data to interactive data app in an hour