Updating Late window data

Hello All,

I have a scenario in which i am getting records from users like -:


So the records are coming in a stream which i am getting from kafka.

The scenario is like 1st record i get is like-:


again if the record is updated then it is pushed again in kafka and i get-:


I am getting records from kafka and pushing the records into druid and using superset to visualize realtime data.

So what i need to do is i need to count the records on the basis of stages and if a particular record moves from 1 stage to other then that stage should reflect 1 less record and other stage would be reflecting 1 more record.

I checked out druid for data updation and could not find anything related to my scenario.

Kindly help on this.