spark aggregations - using one aggregation result as in input to another aggregation (within the same groupby)

Question

I have a huge dataset (billions of rows) that summarize user behavior. For e.g. event type and the number of times a user performed that action

Sample data looks like thi