Fisheries forecasting, physical approach comparison between regression and autoregressive integrated moving average (ARIMA)

Abstract

Empowerment of fishery resources requires analysis of forecasting results as an effort to maintain sustainability and human needs. Forecasting is an approach to predicting based on past facts, which is expected to be used as a decision support system. However, there are problems with the accuracy. In this study, we compare the regression method with ARIMA to find out which method can be used as the most appropriate choice in fisheries forecasting with physical benchmarks (seasonal and climate). We conduct a systematic literature study on various studies with the theme of fisheries forecasting. The search focuses on studies with the main criteria in the form of an explicit discussion of the basic forecasting methods and literature on marine physical influences. Then define a search method by combining fishery stocks or landings against forecasting with marine physical features optionally using PRISMA. The results show that the ARIMA models have a better fair value and accuracy than the regression because the ARIMA model captures the history of data autocorrelation and extrapolates it to the forecasting framework that will be carried out. Hence, it is most suitable for use with additional marine physical variability.