Developing Spidey Senses
summary
Shows how to detect time-series anomalies from JavaScript using Azure Anomaly Detector (REST + Node SDK app-azure-ai-anomalydetector) with a minimum 12-point history, daily–yearly granularity and a free tier at 10 calls/sec. Demonstrates local data collection on a Raspberry Pi (TSLab + node packages), sends timestamps/values, flags a fabricated 132° outlier vs normal ~32° and lights the Pi as an alert — the API classifies the point as anomalous and the LED response works. Mentions multivariate preview (up to 300 signals) and container/edge deployment.
AI-written summary · about
chapters (13)
- 0:00Start
- 0:45Spidey Sense?
- 2:28Real Spider Sense
- 4:24Anomaly Detection
- 6:19Rule-based Systems
- 7:47Statistical Techniques.
- 12:59Time Series Data
- 14:06Time Series Anomaly Types
- 16:42Azure Cognitive Services
- 18:38Limitations
- 19:18Calling the Anomaly Detector API
- 32:14Where can you use this?
- 33:14Metrics Advisor
related talks
32mDeveloping "spidey sense"
28mReal Time Sensor Anomaly Detection with Sci Kit Learn and the Azure Stack
52mDevil in the Haystack
51mAnomaly Detection as a Service - What is normal?
35mObservability’s Sixth Sense: Detecting Anomalies in Metrics
19mEnterprise AI-Powered Predictive Maintenance Platform
18mCross-Industry Anomaly Detection Solutions with R and Shiny
29mConnecting the Dots: Unleash the magic of AI in IOT