摘要：The utilization of Generalized Additive Models (GAMs) it is common to predict the fishing ground in Indonesian sea territories with high accuracy. This research aims to predict the fishing ground of yellowfin tuna using GAM. The data used in this research is tuna fishing data from logbook of Bungus Ocean Fishing Port and oceanography data. The analysis of fish abundance was expressed in the value of hook rate tuna longline. Hook rate is the stock density index. In modeling, the dataset was divided into 2 sections: training data used for model formation and evaluation data used to validate predicted results from the model. In this study, 2015 data was used as the training data and data in 2016 was used as the evaluation data. The results showed that, the prediction had generated a total of 14 models through GAMs statistical approach based on oceanographic parameters. Model SST+Salinity+SSH+Chl-a was found the best with the smallest AIC value of 658.1 and the largest deviance value of 56.9%. The deviance value indicated that the GAM equation could explain 56.9% of the hook rate data. According to the GAM model, potential fishing ground in 2016 were on Siberut and Sipora Islands.
其他摘要：Pengunaan Generalized Additive Model (GAM) sudah umum digunakan di beberapa wilayah laut Indonesia dengan tingkat akurasi yang lebih baik. Tujuan dari penelitian adalah untuk memprediksi daerah penangkapan ikan tuna sirip kuning melalui pendekatan statist
关键词：generalized additive model; Indian ocean; remote sensing
其他关键词：generalized additive model; samudera Hindia; penginderaan jauh
