The first thing I thought when I saw the data was "well, we've got to unpivot those columns" and then I looked at the transforms and see that they had just been pivoted. So, remove that step.
You could split the 'Nominal - Dec' to get a column that indicates job role e.g. Graduate.
The 'CY Total' can then become
CY Total = CALCULATE(SUM(TableSalary[Base Amount]), TableSalary[Accounting Period] = "2019/001")
and PY Total can be re-written similarly.
%Increase can stay.
Weighted average would be
Weighted Av = [% Increase] * [CY Total]
You can pull all the relevant fields into a matrix with 'job role' in the column and employee in the rows.
All the Totals and sub-totals will give you the figures you want.
Alternatively, to see the data in more compact form, create two tables 1 - Role, weighted av, % increase
2 - Employee, Role, all the measures
------------------------------
For a more complex data set, you could implement the measures using time intelligence rather than filtering the table with hardcoded years i.e. "2019/001"
Hope that helps.
