Additional models

[dashaub]

We should consider handling additional models. If we implement issue #11 this will need to be considered.
Furthermore, once this package is stable I'd like to work on and include arar/ararma models (see robjhyndman/forecast#8)

bats() models might be an addition. Since they are quite similar to tbats() it may not be worth inclusion, but they can provide better fits with default settings in some causes (e.g. compare bats(taylor) to tbats(taylor)).

[ellisp]

Cross reference #50 and #51 for two other possibilities.