Once enough filters are applied, we can narrow down to specific observations within cubes. We can then show the value of these observations as a preview.
We don't necessarily need to filter everything that applies. Indeed there will be very few situations where multiple cubes have a completely overlapping sets of dimensions (typically filtering all the dimensions for a given cube would mean no other cubes matched).
We can also make some inferences with heuristics e.g.
unless specified, you're probably interested in the latest data
if there's only one possible choice, we can just select (i.e. present) that
if a codelist has a single top concept, then we can start with that
Here the heuristically chosen dimensions/ values are presented in grey (with the rule in brackets). These aren't editable as they're rules not settings, but you can remove them (removing the preview). These ought to update reactively as the explicit filters are applied.
Alternatively, rather than showing all the guessed component-values as further columns in the table, we could hide these assumptions and instead present a single representative figure from the dataset. We could present this with a prose context using templates written for each cube.
Not every observation will be amenable to this method. The labels of some codes - e.g. from the Harmonised System - are paragraphs in themselves!
Extending the description in the mockups.
Once enough filters are applied, we can narrow down to specific observations within cubes. We can then show the value of these observations as a preview.
We don't necessarily need to filter everything that applies. Indeed there will be very few situations where multiple cubes have a completely overlapping sets of dimensions (typically filtering all the dimensions for a given cube would mean no other cubes matched).
We can also make some inferences with heuristics e.g.
Here the heuristically chosen dimensions/ values are presented in grey (with the rule in brackets). These aren't editable as they're rules not settings, but you can remove them (removing the preview). These ought to update reactively as the explicit filters are applied.
Alternatively, rather than showing all the guessed component-values as further columns in the table, we could hide these assumptions and instead present a single representative figure from the dataset. We could present this with a prose context using templates written for each cube.
Not every observation will be amenable to this method. The labels of some codes - e.g. from the Harmonised System - are paragraphs in themselves!