This study is based on logbook data from longline fisheries targeting albacore tuna (Thunnus alalunga) in the South Pacific from 2020 to 2022. A geographically weighted random forest (GWRF) model was applied and Shapley additive explanations (SHAP) were incorporated to develop an interpretable habitat prediction model and analyze the influence of key environmental factors on albacore tuna distribution. The study aims to provide a scientific basis for habitat research and the sustainable management of albacore tuna in the South Pacific. The results indicate that compared to the traditional random forest (RF) model, the GWRF model improves key performance metrics, including precision, accuracy, recall, and the area under the receiver operating characteristic curve (AUC), by 5%–10%. Feature importance and SHAP contribution analyses identified sea surface temperature, sea surface dissolved oxygen concentration, temperature at 50 m depth, and dissolved oxygen concentration at 50 m depth as the key environmental factors influencing albacore tuna habitat distribution. SHAP interpretability analysis further revealed the optimal habitat conditions, indicating that the most suitable habitats were located in areas where sea surface temperature and temperature at 50 m depth ranged from 15 ℃ to 20 ℃ while sea surface dissolved oxygen concentration and dissolved oxygen concentration at 50 m depth ranged from 240 to 260 mmol/m³. Individual prediction SHAP value decomposition further confirmed that suitable temperature and sufficient dissolved oxygen are the key factors influencing albacore tuna habitat selection. These findings provide new insights into the spatial distribution patterns of albacore tuna habitats and the underlying environmental driving mechanisms.