<![CDATA[The western waters of Sumatra in the eastern Indian Ocean are highly productive fishing grounds for yellowfin tuna (Thunnus albacares). However, the dynamic oceanographic conditions of these waters pose challenges for reliable fisheries prediction. Therefore, this study aimed to identify key environmental drivers and predict Potential Fishing Zones (PFZs) using GeoArtificial Intelligence (GeoAI)-based machine learning (ML) framework integrating Guided Regularized Random Forest (GRRF) and Random Forest (RF). The predictive models were developed using multi-source oceanographic variables as predictors, including sea surface temperature (SST), chlorophyll-a (Chl-a), salinity, current velocity, and net primary productivity (NPP). The results of correlation analysis showed that there was a strong relationship between Chl-a and NPP (r = 0.87), indicating the central role of primary productivity in tuna habitat dynamics. Both models achieved high predictive accuracy, with GRRF slightly outperforming RF by reducing variable redundancy and improving interpretability. Tuna occurrence was strongly associated with moderate SST (29.2–29.8°C), high NPP (>350 mg C m⁻² day⁻¹), and zones influenced by the South Java Coastal Current (SJCC) and the Indonesian Throughflow (ITF). Spatially, PFZs were concentrated along the Mentawai slope and adjacent offshore waters, particularly during monsoon transition periods. These results showed the effectiveness of GeoAI-based methods for predicting PFZs and supported sustainable tuna fisheries management. Moreover, future studies should incorporate interannual climate variability (ENSO, IOD) and advanced deep learning frameworks to enhance spatiotemporal prediction.]]>
