FPPPY and FPP3 report different Guerrero lambdas for the aus_production "gas" time series #38
tc-git-1
started this conversation in
Error report
Replies: 1 comment 1 reply
-
|
thanks @tc-git-1! this was fixed by Nixtla/coreforecast#97. we will update the book with the newest coreforecast version soon. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
FPPPY and FPP3 report different Guerrero lambdas for the aus_production "gas" time series.
FPP3 reports 0.11, while FPPY reports 0.08.
I also reported this in the Nixtla Slack community at https://nixtlacommunity.slack.com/archives/C031EJJMH46/p1753107018956849
(FPPPY uses coreforecast.scalers.boxcox_lambda to estimate the Guerrero lambda.)
References:
FPPPY, https://otexts.com/fpppy/nbs/03-decomposition.html#mathematical-transformations
FPP3, https://otexts.com/fpp3/transformations.html
Beta Was this translation helpful? Give feedback.
All reactions