Time - varying state space regression model in fisheries

Abstract

Linear regression analysis is usually based on the assumption that underlyingparameters remain constant. However, in reality, this may not hold true particularlyif data is collected over a long period of time. To handle such a situation,time-varying state space regression modelling approach using Kalman filteringtechnique, needs to be employed. In this paper, this methodology is discussedin detail. As an illustration, a comparative study of above two approachesis carried out using all - India data on marine fish production and marine productsexport. It is concluded that the latter approach performs much better thanthe former