I was recently browsing the machine learning functionality at http://reference.wolfram.com/language/ref/TimeSeriesModelFit.html
I think adding models with exogenous variables would be very useful i.e. like ARMAX . Sometimes the time series does not exist in isolation. For example if we are forecasting the electricity price one of the most important variable is not its own series but a series of temperatures alongside it. Thank you very much!
