Meta-data scope of the package: Progressive aggregation

[abenassi]

When facing hierarchical groups of time data series, there are groups of them that added together conform an aggregated time data series that is also present in the spreadsheet. In these cases, the name of the low-level series should be composed to reflect that they are part of a high-level one (eg. "GDP" can be decomposed into 10 chapters like "Agriculture", "Services", etc. - see test cases 6 and 7 for an example of this). This could be addressed later on the process, as something optional, after the package already extracted successfully all the series with their low-level names. The implementation of this feature though, is a low priority for the moment given its complexity and little value added for the expected user case.

• Exploring all the possible adding combinations of the first values of the series in order, assuming that aggregated series should immediately follow (or being right before) a group of low-level ones (test case 6) could be a reasonable strategy.
• Exploring all the possible adding combinations disregarding the order is another possibility, but this would take much more time.
• Using additional characteristics of the header cells format could be another strategy, but is complicated to generalize enough to make this strategy useful in a variety of cases (test case 7).