-1

I'm currently looking for particular service in Azure for my side project.

I'd like to build a real-time service that gives alerts or notifications to users by detecting anomaly values from a time series data of temperatures collected over a time period of 10 months(More than 5,000,000 data exist in Azure Cloud)taken from a few devices.

I've tried Anomaly Detector API, and it gives me detected values that are different from the expected values by the model, but since there is a limited input of 8,640 time series points, I can't be sure it is the model or method that I'm looking for.

My goal is to be able to input real-time temperature data through IoT sensors from few devices, like every 5 seconds, to the model, into the model that detects anomaly values based on patterns that the model would have extracted.

Thanks a lot in advance!

1
  • 1
    Is there a reason you can't build it yourself? 5,000,000 data-points isn't much - you could even probably do it all in SQL...
    – Dai
    Commented Jul 1, 2022 at 3:10

1 Answer 1

0

There are few services in Azure which allows you to implement Anomaly Detection. Those services are:

  • Azure Stream Analytics
  • Azure Data Explorer
  • Azure IoT Edge

To implement the Anomaly Detection in Data Explorer, you can refer Microsoft official document - Anomaly detection and forecasting in Azure Data Explorer

The same can be implemented using machine learning in stream analytics service. Refer Anomaly detection in Azure Stream Analytics .

The same ask in Microsoft Q&A helps you to integrate Azure IoT Edge and Stream Analytics for Anomaly Detection. Source Microsoft Q&A - How to use Azure Anomaly Detection on IOT edge module and Author SandervandeVelde42

Not the answer you're looking for? Browse other questions tagged or ask your own question.