Many time series show multiplicative effects, such as seasonality that changes as the underlying level changes. These are easily captured in GAMs using a tensor product (i.e. te(time, season)), but this strategy suffers from poor extrapolations of the marginal time basis. For identity links (gaussian, lognormal, student-t), it may be useful to allow effects to be multiplied by the trend estimate
Many time series show multiplicative effects, such as seasonality that changes as the underlying level changes. These are easily captured in GAMs using a tensor product (i.e.
te(time, season)
), but this strategy suffers from poor extrapolations of the marginaltime
basis. For identity links (gaussian, lognormal, student-t), it may be useful to allow effects to be multiplied by the trend estimate