Predictive Website Dashboard for Coral Thermal Stress in Indonesia - XGBoost Models and Training Data

Key Points

• This research aims to develop a forecasting system to predict coral thermal stress based on XGBoost models.
• Utilized processed training datasets from 5 reef locations covering 1986 to 2025.
• Implemented XGBoost models trained with 5-fold cross-validation evaluation metrics.
• Included complete reproducible code for data collection, preprocessing, and model training.
• Effectiveness of XGBoost in predicting coral thermal stress at various reef locations was documented.
• Cross-validation metrics demonstrated reliable performance of the forecasting system.
• Reproducible code facilitates further research and application in coral monitoring.