Open 03rcooke opened 3 years ago
I've been looking into this and I think I can explain. The reason is that the whole model is based on estimating growth rates between adjacent years. This means that any uncertainty in the input data for year 1 gets propagated into the growth rate between years 1 and 2, not the index values. NB M.prime will have uncertainty in year 1 if there are species with missing data for that year, which explains why you get uncertainty for a baseline of 1970 but not 1994. Is this something that needs further documentation?
I don't fully understand the parameters of the bma indicator, but as far as I can tell all combinations of the parameters produce zero error in year 1 for this test data.
Which is strange as sometimes I seem to produce non-zero error in year 1 with actual data (when I filter the data to 1994-2020, but not when I use 1970??)
Reproducible example:
@drnickisaac