For predictSolute.loadReg2, currently the only control over this is the allow.incomplete argument to rloadest::predLoad, which is just a logical:
If allow.incomplete is TRUE, then loads will be computed based on all nonmissing values, otherwise
missing values NAs will be returned. For this application, missing values includes NAs and gaps in
the record, except for by set to "total" or user defined groups where missing values only includes
NAs. For prediction by "day" when there are variable number of unit values per day, allow.incomplete
must be set to TRUE.
It might be nice in the future for users to specify a threshold for what is deemed 'complete'. There are probably a couple ways this could be done, with varying levels of complexity for different time periods. Challenges would include knowing what data should be there for a complete period, handling different combinations of periods/data frequency, and possibly edge cases with first/last periods. The argument would need to either be a fraction, or perhaps it would be easiest to just be number of measures, relying on the user to know what they want for a given aggregation period.
For
predictSolute.loadReg2, currently the only control over this is theallow.incompleteargument torloadest::predLoad, which is just a logical:It might be nice in the future for users to specify a threshold for what is deemed 'complete'. There are probably a couple ways this could be done, with varying levels of complexity for different time periods. Challenges would include knowing what data should be there for a complete period, handling different combinations of periods/data frequency, and possibly edge cases with first/last periods. The argument would need to either be a fraction, or perhaps it would be easiest to just be number of measures, relying on the user to know what they want for a given aggregation period.