Detect anomalies in manufacturing data using Amazon SageMaker Canvas

This article demonstrates how to use Amazon SageMaker Canvas to detect anomalies in manufacturing data, without requiring coding expertise. It shows how a domain expert can curate data, select relevant features, train a regression model to predict normal behavior, and then deploy the model for anomaly detection.

Specifically, the article covers:

• Overview of anomaly detection use case in manufacturing
• Using SageMaker Canvas to curate data, select features, train regression model
• Deploying the model as an endpoint
• Using the deployed model with Lambda to detect anomalies by comparing predictions to actual values
• Interpreting anomaly scores and setting thresholds
• Clean up steps