forecastxgb-r-package
forecastxgb-r-package copied to clipboard
thorough tests for the xreg functionality
not sure how to do this because I don't know of a large scale testing suite of data equivalent to Mcomp with explanatory variables
Simulated data is always a possibility, although it has plenty of limitations.
Rob Hyndman doesn't know of any data collections of the sort I'm looking for so I guess I will need to do simulations. They won't be great for real performance tests (eg xgbar v. arima) because they'll just favour the method that most resembles the simulation. But they'll give some idea and should at least surface stability or other bugs and problems.