Times series data

[go-bayes]

Is it possible to specify a group-level factor? For example, imagine repeated measures with missing observations. Thanks for all the work in this package.

edit I see you cover this on p.32 -- the answer is "no but... " -- would be great to see this as a feature if possible, one day...

[tsrobinson]

Hi @go-bayes -- thanks for raising this issue. I'll keep it open because it's definitely something we are actively considering adding to MIDAS.

One solution we are toying with here is to add recurrent neural network cells that would allow us to model states, and therefore explicitly model sequences/time series data.

[go-bayes]

Awesome, you folks are great -- really grateful for all your work here -- thank you!