I loved the talk. I am very interested in hearing more about making Fable reconciliation on par with HTS in terms of computational efficiency. Also, I would love a non-negative constraint method added to the Fable reconciliation function. Does anyone have a link to the non-Negative Cross Temporal Forecast Reconciliation talk mentioned at the end of this video?
There was a talk given at ISF 2022 entitled Tools for forecast reconciliation: the R package FoReco by Daniele Girolimetto which might assist with your inquiry. I found this reference as well: cran.r-project.org/web/packages/FoReco/FoReco.pdf.
What a gem, professor Hyndman is the MVP
This is a great talk! Thanks for posting this
I loved the talk. I am very interested in hearing more about making Fable reconciliation on par with HTS in terms of computational efficiency. Also, I would love a non-negative constraint method added to the Fable reconciliation function. Does anyone have a link to the non-Negative Cross Temporal Forecast Reconciliation talk mentioned at the end of this video?
There was a talk given at ISF 2022 entitled Tools for forecast reconciliation: the R package FoReco by Daniele Girolimetto which might assist with your inquiry. I found this reference as well: cran.r-project.org/web/packages/FoReco/FoReco.pdf.
What’s the difference between matrix S and S’. Is it S_transpose
Yes!