TITLE:
Empirical Modeling of Annual Fishery Landings
AUTHORS:
Eric Adjei Lawer
KEYWORDS:
Artificial Neural Network, ARIMA, Exponential Smoothing, Forecasting, Fish, Ghana
JOURNAL NAME:
Natural Resources,
Vol.7 No.4,
April
21,
2016
ABSTRACT: Forecasting plays an essential role in policy formulation and implementation especially in the management of fisheries resources. In this paper, various techniques of forecasting using time series analysis were evaluated on annual fishery production data. In addition to the Box-Jenkins approach, other methods such as the feed forward neural network and exponential smoothing approaches were also examined. A parsimonious model for each forecasting approach was then selected using penalized likelihoods. The chosen models were then evaluated based on their ability to produce accurate forecasts. Implications of the findings as discussed revealed that no particular method was ideal for modeling all landings. Hence when forecasting fishery landings, it is recommended that different structural approaches be compared before selecting an appropriate one for use.