Review the calculation workflow for generalization

[mgdenno]

We need an evaluation of the workflow we use to perform calculations on the joined timeseries. Can we generalize it to be more generic and more extensible?

• Pre "group by and aggregate" (i.e. attributes and user defined fields), but this could also be a step in the processing if there was/is a need.
• Filter (remove rows from the population)
• Group by (determine the populations to be evaluated)
• Aggregate (reduce the rows to those created by the groups and add any aggregates, mean, bias, kge, etc),
  • This will be a Python function and could include an additional filter or sub-aggregates (i.e. calculation of mean daily)
  • This can return a single value of a struct or list. If struct or list it may require unpacking.
• Finalize (A few things can happen here. Multiple aggregated values can be combined, an aggregate the creates a Struct can be unpacked, what else?