Aggregation expressions are determined by the following properties:
Suppose that the time series database contains N identifying dimensions, and the aggregation filter has M of them set.
When the element is set, it is allowed to use N, N+1, N-M or N-M+1 parts in the expression:
On specifying N parts it is considered that the user specified all dimensions, and for this element the filter is not used. Example: {Albania|BCA} or {{Country}|BCA} (with parameter).
When N+1 parts is specified, it is considered that the user specified all dimensions + calendar level. For this element the filter is not used. Example: {Albania|BCA|Q}.
When N-M parts are specified, it is considered that the user specified only the dimensions, which are not in the filter. Example: {BCA}, filter contains Country dimension.
When N-M+1 parts are specified, it is considered that the user specified only the dimensions, which are not in the filter + calendar level. Example: {BCA|Q}, filter contains Country dimension.
If M=1, and N parts are specified in the element, the attempt of "a" variant analysis is made. If it is unsuccessful, the attempt of "d" variant analysis is made. There is a pitfall here: if the last specified dimension contains A, S, Q, M, W, D elements, it is impossible to specify calendar dimension because the analysis is executed using the "a" variant.
In other cases an error occurs.
If N = M, the X symbol is used to set the element from the filter.
If the analysis is not used (the element selection is set manually) to set aggregation expression, use the following rules:
By dimensions, missing in the aggregation filter, the single selection must be set.
For the dimensions, which are in the filter:
Empty selection is specified. On calculation the selection from the filter is used.
The single selection is used. It is considered that the user specified all dimensions, and for this element the filter is not used.
See also: