Using the multivariate Azure Anomaly Detector with ThingWorx
With ThingWorx, we can already use univariate anomaly alerts (on a single sensor value). However, in many situations, the readings from an individual sensor may not tell you much about the overall issue and a multivariate anomaly detector can be more useful.
This post is intended to provide an overview of the Azure Anomaly Detector and how it can be integrated with ThingWorx. The attachment contains:
A document with detailed instructions about the setup;
A .csv file with the multivariate timeseries dataset;
A .twx file with some entities that need to be imported in ThingWorx as well as the CSVParser extension that needs to be installed;
A .zip file that will need to uploaded in an Azure Blob Container at some point in the setup