Not sure how clearly I can communicate this, but let me try.
Say I have a dataset for sale data with these columns.
SalesPerson,SalesMonth,SalesRegion,SalesAmount.
From this dataset, I want to make a matrix report. The row groups will have SalesMonth and SalePerson.
Now, about the column group... this is where my question is.
Normally, I would just drag and drop SalesRegion for the columns groups and throw in SUM(SalesAmount) at the intersection and I would have a typical matrix report with SalesMonth-SalesPerson for the rows and SalesRegion for the columns.
But instead of showing the SalesRegion as they are in the dataset, I want to have certain custom groups for column groups. Now, I know I can add another column in the dataset that has these group values and use that column for the column group. But the tricky part is that a SalesRegion can fall into multiple groups.
For example, let's say there are 3 regions, RegionA, RegionB, RegionC.
For my column groups, let's say I want to show Group1 and Group2. Group1 is the sum of Region A and Region B. Group2 is sum of Region A and Region C.
Can you do this with the given dataset with some creative use of filters. Or would it involve creating two calculated fields? One to mark if a region is in group1 or not and another field to mark if a region is in group2 or not and using those fields and adjacent columns groups? I tried this, but I am not able to filter out the corresponding sums.