search

BiologicalRecordsCentre / BRCindicators

An R package for creating indicators from trends data
4 stars 11 forks source link

Separate fitting from rescaling in bma() #5

Open drnickisaac opened 6 years ago

drnickisaac commented 6 years ago

At present, bma() fits a model in JAGS then rescales the output. These should be separated into two different functions. The first function should return the JAGS model object. Ideally, the second function should be fully aligned with (and duplicate?) the existing rescale_indicator() function.

In practice, it may be best to achieve this by converting bma() into a wrapper with two components, of which one is an updated rescale_indicator()

AugustT commented 6 years ago

park this until the other methods are ironed out

drnickisaac commented 4 years ago

I did some tests recently that showed rescaling does make a big difference if there are many missing data near the start of the series. To fix this I need to get my head around what the rescaleYr argument is doing (see #4) and the rescale_bayesian_indicator function within bma()