Structural time -series modelling for describing fish production

Abstract

Statistical modelling of marine, inland, and total, fish production of the country for the period of 1980-81 (0 1997-98 was carried out using' Local linear trend model' (LLTM), which belongs to the family of 'Structural time-series models'. The model was first put in state space form and subsequently Ka'lman filtering technique was applied for estimation of parameters. Results obtained through thisimethodology were compared with corresponding analogue from the family of ARIMA time-series models. It was found that the former approach performs better than the latter. Finally, the identified model was used to forecast marine, inland, and total fish production for the next 5 years.