SARIMA MODELS WITH MULTIPLE SEASONALITY

SARIMA models and exponential smoothing methods are classical ap- proaches to account seasonal dynamics. However, they tipically allow to model just one periodic component, while many empirical time series data show multiple season- ality, possibly interlacing toghether. To face this case, different decomposition models have been proposed in literature, while SARIMA models have been quite neglected. To fill the gap, in this work we suggest a suitable specification of the SARIMA model, called mSARIMA, able to account multiple seasonality. To study its performance, we compare it with two popular seasonal-trend decomposition approaches, namely the TBATS and MSTL models. A simulation exercise shows that mSARIMA models are more effective in describing the the different seasonal components.