As the STEP_COMPLETED_DATE is a filter, it will reduce the dataset. One way that I can think of to always have the dataset static irrespective of the filter is to do something like following but on a query level :-
1) Create a standalone table with all the distinct STEP_COMPLETED_DATE values
2) FULL OUTER JOIN the main dataset on the standalone table on STEP_COMPLETED_DATE.For matching dates, keep the existing measure value and for non matching default to 0.
The above solution will be okay if the volume of main dataset is not huge.
PS - Even I am interested to know how to keep the dataset static irrespective of the filter