Detect anomaly for string parameters using Azure Stream Analytics

Question

I am working on anomaly detection using Azure Stream Analytics. I want to detect anomalies in request payloads passed to a Rest Api.

Azure Stream Analytics provides a built in function AnomalyDetection_SpikeAndDip, for detecting anomalies in the data.

The signature of the function is as follows:

AnomalyDetection_SpikeAndDip(<scalar_expression>,<confidence>,<historySize>,mode>)

This function only accepts one scalar expression(can be integer or float field in from the request payload).

Now, my request payload contains fields of type strings as well. Is there any way that we can detect anomalies in fields of type strings as well?

Answer 1

Built in anomaly detection functions in Stream Analytics doesn't accept strings. The alternate option is to use Azure Machine learning to build your custom model and integrate with Stream Analytics.