Multiscale functional autoregressive model for monthly sardines catches forecasting

In this paper, we use a functional autoregressive (FAR) model combined with multi-scale stationary wavelet decomposition technique for one-month-ahead monthly sardine catches forecasting in northern area of Chile (18 ^o 21S∈-∈24 ^o S).The monthly sardine catches data were collected from the database of the National Marine Fisheries Service for the period between 1 January 1973 and 30 December 2007. The proposed forecasting strategy is to decompose the raw sardine catches data set into trend component and residual component by using multi-scale stationary wavelet transform. In wavelet domain, the trend component and residual component are predicted by use a linear autoregressive model and FAR model; respectively. Hence, proposed forecaster is the co-addition of two predicted components. We find that the proposed forecasting method achieves a 99% of the explained variance with a reduced parsimonious and high accuracy. Besides, is showed that the wavelet-autoregressive forecaster is more accurate and performs better than both multilayer perceptron neural network model and FAR model.