Models Log -lir near y(t)≈K t logy(t)≈B1t+B2 Gaussians ,(t)=e 2 11 An y(1)≈B1e +...+BeModels t y t Ke  ( )  Log-linear: 1 2 log y(t)   t   Gaussians: 2 ( )          j j t j t e    2 2 1 1 ( ) ... 1                    n n t n t y t e e      