search

MicrosoftDocs / azure-docs

Open source documentation of Microsoft Azure
https://docs.microsoft.com/azure
Creative Commons Attribution 4.0 International
10.2k stars 21.34k forks source link

Last Point Anormaly detection #55867

Closed luistelmocosta closed 4 years ago

luistelmocosta commented 4 years ago

Hello, I have a doubt regarding the last point anormaly detection feature. Imagine that I want to detect if the last data point is anormal but I want that prediction to be consistent with the history. Shouldn't the model be trained in the whole dataset first, get the parameters and then evaluate each data point accordingly?

For what I read, a data point is just sent to the API and it is evaluated as anormal or not, how is it possible without any context?

Kind regards

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

GiftA-MSFT commented 4 years ago

@luistelmocosta we will review your feedback and get back to you shortly. Thanks.

GiftA-MSFT commented 4 years ago

@luistelmocosta Latest point anomaly detection is suitable for streaming/real-time data. You send new data points as they are generated to create a model, and the API determines if the latest point in the time series is an anomaly. Meanwhile, with batch anomaly detection, the API will generate a model using the entire time series data and analyze each data point.

GiftA-MSFT commented 4 years ago

I will proceed to close this thread but feel free to comment if you would like the thread to be re-opened. Thanks.