calculate roll up aggregations using dynamic window in spark

Question

I have a spark dataframe which contains month and year. Now I want to use this dataframe to perform roll up aggregations for each month and year for the past months data pre