Matrix drill-down > subtotal from multiple columns

Hey,

I have a problem showing the correct subtotals in a matrix when drilling down to the next level.

What I have

2 datasets:

• Tasks, containing - among others - the following columns:
  • Task ID (link to other dataset)
  • Task Name
  • Task Budget
• Sub-tasks, containing - among others - the following columns:
  • Sub-task ID
  • Sub-task name
  • Sub-task budget
  • Task ID (link to other dataset)

Now, what I want to do is show a matrix listing all the Task names as rows, and their respective data (e.g. budget).

As second row level in the matrix I use Sub-tasks. In each row I want to show several metrics, mostly based on the budget, like budget left, required per day, etc. 

In my current setup however, I use the sum of Task budget as my leading budget amount. So, firstly the Task budget is carried over when I expand to the next level, like in the figure below:

Now the sub-task budget sometimes has a value, and sometimes doesn't. So the budget has to be calculated conditionally I guess.

Ideally, I want to have:

• the task budget always shown (when expanded it is shown as subtotal)
• the sub-task budget shown when it has a value in the dataset
• the sub-task budget not show any thing when it has no value in the dataset.

Can anybody help me achieve this?