Time Traveling in the Cloud
summary
Uses an open NASA engine telemetry set and a docker simulator that spins up 100 simulated engines sending 27-parameter readings into IoT Hub; ingests straight into Time Series Insights for zero‑code exploration, streams into Azure Data Explorer (Kusto) for complex time‑series decomposition and forecasting, and trains an LSTM in Azure Machine Learning endpoint that stream analytics calls to flag maintenance. A window length of ~30 yields about 93% accuracy; hosting runs on ACI and outputs events for downstream Logic Apps.
AI-written summary · about
chapters (27)
- 0:00Introduction
- 0:30About Sam
- 0:54About the session
- 4:18Time Series Components
- 6:01Scenario
- 6:47IOT Hub
- 7:50Reference Architecture
- 9:00Demo Setup
- 10:10Time Series Insights
- 13:28Types
- 14:14Azure Portal
- 15:02Azure Event Hub
- 16:52Overview Portal
- 18:00Zooming in
- 19:15Adding a marker
- 19:38Time shift
- 19:52Modeling
- 20:57Calculated Field
- 22:09Sharing Views
- 23:18Query API
- 24:33Azure Data Explorer
- 33:33Column DB
- 34:03Predictive Maintenance
- 42:43ARgus
- 44:41Azure Machine Learning
- 49:31Stream Analytics
- 55:56Wrap up
related talks
27mLeverage Azure Cosmos DB for time-series for IoT Apps
27mLeverage Azure Cosmos DB for time-series for IoT Applications
19mEnterprise AI-Powered Predictive Maintenance Platform
14mBuilding the future of time series data
01h 01mCreate the Internet of Your Things - The Microsoft Azure IoT Suite
10mBuilding IoT Apps with JSONB & Timeseries
01h 00mIntelligent Interfaces with Cognitive APIs, BOTs and IOT
14mBuilding the future of time series data using Azure Cosmos DB