How to group events by an event variable/property in Aggregate Series API (Gen2) ?

Question

I am currently investigating on how to migrate my TSI Gen1 queries to TSI Gen2 queries and I can't find how to use the Aggregate Series API of Gen2 to do the same thing I was doing with Aggregate API in Gen 1.

Let's say we have these raw data events in my TSI Gen2

I want to aggregate these tarrifId by interval of 1h to know the average consumption by hour.

My query will look like this :

{  
    "aggregateSeries": {  
      "timeSeriesId": [  
        "de1e7bce-c248-44b6-a2c4-f63aa660828f",   
      ],  
      "searchSpan": {  
        "from": "2017-12-31T06:00:00Z",  
        "to": "2017-12-31T10:00:00Z"  
      },  
      "interval": "PT1H",  
      "inlineVariables": {  
        "AverageConsumption": {  
          "kind": "aggregate",  
          "filter": null,  
          "aggregation": {  
            "tsx": "avg($event.consumption.Long)"  
          }  
        }  
      },  
      "projectedVariables": [  
        "AverageConsumption"  
      ]  
    }  
  }  

Now if I want to have this this average consumption by hour for each tariff, I cannot see how to modify my query to do that.
There is nothing than allow me to do a sort compute my averageConsumption "group by" tariffId. One thing to note is that tariffId values are not fixed/static values.

In TSI Gen1 I was able to do this "group by" with nested aggregate / dimension json objects.
Am I missing something or is this not possible anymore ?

Answer 1

@Alexandre thanks for your patience.

I have confirmed that this kind of query - computing across ids grouped by some parameter - in Gen1 is what we're calling 'fleet queries' in Gen2. This feature is not available yet, it's something on the roadmap.

I would suggest that you add it as a feature on UserVoice as well so others can vote for prioritization purposes: https://feedback.azure.com/forums/906859-azure-time-series-insights . If you do so, please share here the link for cross-referencing, I will make sure the feedback is taken into consideration by the team :).

Finally if you can, please share with us the following to help us better understand your specific scenario:

• How many sensors will you be aggregating over? on the scale of 10s? 100s?

Thank you!