Two things:
- where do you want the data loaded to?
- how to achieve this?
Typically when you introduce a lower level you want to apportion the departmet over many committees. When you re-consolidate the committees you hope to have the same value in department as you started with.
As a first step write an export procedure and export your data to a flat file. Now you are backed up so if you make a mistake you can recover.
Next change your dimension by adding committees. For each department add one extra commitee, and name it to reflect the departmet it belongs to. e.g. for Department1 create a committee called DummyDepartment1
Now load the flat file using a synonym table to map the data into the dummy committees.Department1's data will get loaded to DummyDepartment1.
Write a procedure to apportion the data from the dummy committees to the real ones. You will need to decide the business rule to drive this allocation. In a sinple case each department might have 4 committees so you allocate 25% of DummyDepartment1's data to each committee.In real life you probably have a more complex allocation.
Now the data is allocated write a procedure to clear the data from the dummy committees.
Finally delete the dummy committees from the dimension.