A Forecasting Model of WCPO Katsuwonus Pelamis Purse Seine Catch Based on Rough-SVR
Ye-Bo Yang, Hongchun Yuan, Ying Chen · 2007
Fishery prediction by using long-term accumulated hydrology elements and catch statistic data has been an urgent requirement in aquatic domain.In this paper, 18 different WCPO (Western and Central Pacific Ocean) hydrology elements were collected and a key influence element set was found by using Rough Set Theory before training to build up the forecasting model rather than using ε-SVR directly.Comparative experiments with traditional ε -SVR show that the Rough Set could remove redundancy elements effectively; the Rough-SVR results in a better goodness of fit than the traditional ε-SVR and it is superior to multiple regression analysis.