Performing Multiscale Autoregressive (MAR) Order 1, 2 And 3 Modeling for Hairtail Fish Production Forecast

Full Text PDF PDF
Author(s) Handayu Putrindi | Henny Pramoedyo | Agus Widodo
Pages 140-143
Volume 5
Issue 4
Date April, 2016
Keywords Wavelet, Multiscale Autoregressive (MAR), Fish Forecast, MODWT
Abstract

Hairtail fish is one of high demand export commodities for fish in Brondong Fishing Port. The practices of large-scale fishing performed by local fishermen in the last few years have caused the decrease in hairtail fish production per unit of fishing rod, which is regarded as the standard fishing gear for export purposes. Therefore, the production forecast of hairtail fish for one year ahead will be necessarily helpful for its resource management planning. Regarding that purpose, this research endorsed a wavelet transformation to predict the hairtail fish production by fashioning the model of Multiscale Autoregressive (MAR) in which the predictor was obtained from the Maximal Overlap Discrete Wavelet Transform (MODWT) decomposition using wavelet Haar filter. MAR order was determined based on the autoregressive order, through the ARIMA process. Yet, this hairtail fish production data is a moving average time series that more than one MAR was required. This research attempted the use of order 1, 2, and 3 on MAR modeling. Several selected MAR modeling were such models equipped with assumptions of normality and white-noise on residual. The wavelet forecast model involved the best MAR modeling build upon the squared correlation (R2) closest to one. The purpose of this research is to form the best wavelet forecast model for Hairtail fish production data in Brondong fishing port, and forecast the Hairtail fish production for the next 12 month.

< Back to April Issue