search

olshena / COVIDNearTerm

1 stars 0 forks source link

Harmonizing functions, part 1 #9

Closed olshena closed 4 years ago

olshena commented 4 years ago

For the ETS model, we have a single build and predict function. For the autoregressive model, we have separate build and predict functions. Should we have separate build and predict, or a single one, or should it be separate between the two? Let's discuss.

kikapp commented 4 years ago

I think it makes sense to have as much internal as possible while still allowing users to customize. Having separate functions for buildAR and predictAR will hypothetically give users more options for generating input for predictAR. I'm not sure that will be all that useful. However, maybe we can create a third function that serves as a wrapper around both?

olshena commented 4 years ago

I think a wrapper makes sense. Should the wrapper also be able to call simulateHoltMA, or just for AR?

kikapp commented 4 years ago

I think it makes sense to keep it just for AR. These two sets of functions are doing slightly different things and users should hopefully be able to pick the right one for the job

olshena commented 4 years ago

I now agree. Perhaps it should be called simulateAR, or something like that, to echo the other function.

kikapp commented 4 years ago

Created a simulateAR function which combines buildAR and predictAR