I believe there is an incorrect reference to the sigma attribute of NNETAR in generate.NNETAR(). It is referred to as sigma
When it reality I believe it should be sigma2, per the output of a NNETAR model:
library(fable)
#> Warning: package 'fable' was built under R version 3.6.2
#> Loading required package: fabletools
#> Warning: package 'fabletools' was built under R version 3.6.2
library(fabletools)
library(tsibbledata)
#> Warning: package 'tsibbledata' was built under R version 3.6.2
library(tsibble)
#> Warning: package 'tsibble' was built under R version 3.6.2
fit <- as_tsibble(airmiles) %>%
model(nn = NNETAR(box_cox(value, 0.15)))
fit$nn[[1]]$fit$fit
#> # A tibble: 1 x 1
#> sigma2
#> <dbl>
#> 1 0.211
Created on 2020-08-18 by the reprex package (v0.3.0)
There error I was getting in my "production" dataset that lead me to this was Unknown or uninitialised column: 'sigma'., but I haven't been able to fully reproduce it because it requires passing through a series of if statements in train_nnetar, forecast.NNETAR, and generate.NNETAR that I haven't been able to replicate with reprex and package data from tsibbledata.
I believe there is an incorrect reference to the sigma attribute of
NNETARingenerate.NNETAR(). It is referred to assigmafable/R/nnetar.R
Line 358 in 9ee1d6d
When it reality I believe it should be
sigma2, per the output of aNNETARmodel:Created on 2020-08-18 by the reprex package (v0.3.0)
There error I was getting in my "production" dataset that lead me to this was
Unknown or uninitialised column: 'sigma'., but I haven't been able to fully reproduce it because it requires passing through a series ofifstatements intrain_nnetar,forecast.NNETAR, andgenerate.NNETARthat I haven't been able to replicate with reprex and package data from tsibbledata.