Semi-additive measures: AverageNonEmpty, LastNonEmpty
It is very helpful that Jedox has implemented the first semi-additive measures in version 6.
(AVG, Last/First Child)
It would be great if this could be continued by 2 more of them:
Most important would be AverageNonEmpty, to get an avarage only across children, who have a real value.
Another one would be LastNonEmpty, which should give you in a certain time period the last real value.
.. if you monitor price changes or inventories, it should give you the last price or the last stock value for each product in a year
(independent if this value is located in Jan, Feb, March or whatever month).
..a further example would be:
get the aggregated volume of the last purchase of each customer.
(The point of time of the last purchase is probably a different one for each customer.
One customer did his last purchase just 1 month ago, another might have had the last purchase 9 month ago or longer)
NonEmpty related to the current Jedox version means actually Non-Zero, as there is no real empty state for cells currently.
But if "empty" will be implemented in future versions, then the aggregation should be real NonEmpty.
AdminJedox (Admin, Jedox AG) commented
since the recent request partially overlaps with a previous request for the semi-additive calculations, we have merged the two suggestions.
If you think that support for a mixture of B- and C-target in rules has value apart from semi-additive calculations, you can of course create a new request for this.
Jens Lorentz commented
It would be great to define rules, which are C-rule related to the time dimension but at the same time B-rule to all other dimensions in the cube.
(So we target consolidated elements in time dimension and base elements in all other dimensions)
With this we could define advanced semi-additive aggregation behaviour, which is currently not provided by Jedox (e.g. NonEmptyAverage)
Right now we only have some basic semi-additive aggregation in Jedox.
In general it would be really powerful to define rules, which are targeting consolidated elements in some selected dimensions and base elements in all other dimensions.
(not only related to time aggregation)