Hello everyone, I'm trying to solve a problem familiar to most universities: forecasting heads and credits.
I have a hierarchical time series forecast using the hts package to forecast heads, but now I want to add in credits as well. Hyndman's grouped time series package, gts (https://otexts.com/fpp2/gts.html#eq:middlelevelXY) seems to be for when two or more time series share the same unit of measurement, in this case the prisoners dataset in the fpp2 package. Credits will always be a multiple of heads, but the exact multiple will change over time and over years.
Currently I forecast the whole funnel of enrolled heads with a bottom level of time series like this:
enrolled, deposited-but-not-enrolled, admitted-but-not-deposited, etc
My question is how can I add in credits to be forecasted simultaneously? For more discussion on my code see: Hierarchical Time Series Forecasts Suggestions for Improvements
Thank you in advance