DataEngBytes 2023 - PER-T1-07 - Nadia Reyhani
summary
Demonstrates running ML entirely inside Amazon Redshift Serverless using Redshift ML to call SageMaker Autopilot, or with Auto=off to train an XGBoost model in SageMaker and import it as a SQL function; supports local inference (bring-your-own model) and remote inference (SageMaker endpoint). Shows a churn demo using a public bank dataset, training cost about USD 11 for ~1.5 hours, and gives concrete model results (one model misclassifies ~254 churners, a poor imported model loses ~4,700).
AI-written summary · about
chapters (13)
- 0:00ML in data warehouses
- 2:08Implementing ML in warehouses
- 4:38Redshift Serverless & ML overview
- 7:29Redshift ML terms
- 8:33Customer churn use case
- 10:18Demo setup & data inspection
- 12:24AutoML in Redshift ML
- 14:50Model training & evaluation
- 17:36Custom model training
- 20:28Bringing models into Redshift
- 23:15SageMaker remote inference
- 25:22Cloud provider alternatives
- 27:05Resources & closing
related talks
52mGenerate ML predictions without writing any code
32mBuild quality ML models easily & quickly with Amazon SageMaker Autopilot
52mPreparing data for reporting, ML, and AI on AWS
19mUsing Amazon SageMaker to develop and deploy ML models
27mScaling ML to the next level: Hosting thousands of models on SageMaker
33mBuild, train, and deploy ML models using Amazon SageMaker
44mPragmatic Machine Learning in the Cloud
39mTrain ML models at scale with Amazon SageMaker Training