Announcing General Availability of Multivariate Anomaly Detection

This post has been republished via RSS; it originally appeared at: New blog articles in Microsoft Community Hub.

Today, we are thrilled to announce that Multivariate Anomaly Detection is Generally Available!

 

Multivariate Anomaly Detection is a new GA feature in Anomaly Detector service, which is a AI service under Cognitive Services that have helped Microsoft partners and customers build metrics monitoring solutions in scenarios like business KPI monitoring, utilization metrics monitoring, predictive maintenance and more. To leverage this new feature, you do not need to know much about machine learning, and you do not need to prepare labels at all for the underlying model is unsupervised. You could easily leverage a set of APIs to build an awesome model on your multiple variables in your system or equipment, and get anomaly detection results with actionable insights!

Louise_Han_1-1667465114296.png

 

Key capabilities and new experience!

 

APIs Introduction

We provide several easy-to-learn and easy-to-use APIs to help you create a Multivariate Anomaly Detection model within minutes and provide two types of inference according to your preference and use cases. You could use the Batch Inference API for a long period of data validation or for a micro-batch task, which works in an asynchronous way together with the Get Inference Results API. If your case is more like real-time or near real-time, the Streaming Inference API might be an excellent choice for you since it will present the anomaly results immediately once you call this API and get every timestamp detected per call.

Louise_Han_2-1667465135307.png

Check the API referecence documentation here! 

 

Quicker and easier data preparation process!

In this GA version (v1.1), we improve the data preparation process after learning customers’ feedback. Now we support two types of schemas for training and batch inference: OneTable and MultiTable. You could use either of these schemas to prepare your data and upload it to Storage Account for further training and inference.

  • OneTable (recommended) is one CSV file that contains all the variables that you want to train a Multivariate Anomaly Detection model and one timestamp column. 
  • MultiTable is multiple CSV files in one file folder, and each CSV file contains only two columns of one variable, with the exact column names of: timestamp and value.

After you prepare your data, all you need to do is upload the file(s) to Storage Account and use the blob URL (which could be easily obtained in Azure Portal) as data path to train a model or trigger a batch inference through multiple ways, like calling the API or use the SDK directly in your Jupyter Notebook, or use in Azure Machine Learning, even Synapse and Azure Databricks are supported as well.

Louise_Han_3-1667465164073.png

 

Customize your own anomaly results!

In this GA version, we add a new field in inference API request body – TopContributorCount. This is a number that you could specify N from 1 to 30, which will give you the details of top N contributed variables in the anomaly results. For example, if you have 100 variables in the model, but you only care about the top five contributed variables in detection results, then you should fill this field with 5, which saves you much time analyzing the results.

 

{ "dataSource": "{{dataSource}}", "topContributorCount": 3, "startTime": "2021-01-02T12:00:00Z", "endTime": "2021-01-03T00:00:00Z" }

 

 

Insightful information in detection results!

Today in the anomaly results you will get:

  1. When there is an anomaly with isAnomaly
  2. How severe is that anomaly with Severity and Score
  3. Contributed variables rank under each anomalous timestamp
  4. Correlation changes among variables that contributed to the anomalous timestamp

Louise_Han_0-1667465610013.png

 

Pricing

During November and December, you could still use this feature (both GA version and preview version) for free. We will update the pricing page on January 1st in 2023 and start charging since then.

 

Customer love around the world

During the public preview stage of Multivariate Anomaly Detection, many customers and Microsoft Partners benefit a lot from this feature. Around 20+ industries have shown their great need for anomaly detection, covering energy, manufacturing, utilities, telecommunications, transport, IT services & business advisory, media, retail and more than these!

Louise_Han_0-1667476652321.png

 

Check out this wonderful customer story with Siemens Healthineers!

“Developing AI-based solutions for detecting anomalies in near-real time during production requires interdisciplinary knowledge from process experts, data scientists, and a DevOps team. Using Azure and its services, we can collaborate seamlessly across domains along the entire machine-learning lifecycle and develop enterprise-ready solutions.”

 

Try Multivariate Anomaly Detection today!

Leave a Reply

Your email address will not be published. Required fields are marked *

*

This site uses Akismet to reduce spam. Learn how your comment data is processed.