Fishing Area Prediction Using Scene-Based Ensemble Models

This study utilized Chlorophyll-a, sea surface temperature (SST), and height (SSH) as the environmental variables to identify skipjack tuna catch hotspots. conducted statistical methods (decision tree, DT, generalized linear model, GLM) ensemble models that were employed for predicting area each time slice. Using spatial historical data, model was trained one of sets. For prediction, correlations new inputs applied select predictive model. scene-based with highest input correlation, this further identified fishing in every case whether alterations their environment affected abundance or not. Overall, performance achieved over 83% correlation coefficients (CC) based on accuracy assessment. concluded DT appears perform better than GLM areas. Moreover, most influential variable construction indicating presence primarily influenced by regional temperature.