PENERAPAN DEEP LEARNING DALAM SISTEM INFORMASI GEOGRAFIS UNTUK ANALISIS DAMPAK PERUBAHAN IKLIM

Climate change is increasingly contributing to the frequency and intensity of natural disasters, especially in archipelagic countries like Indonesia. This study aims to develop a predictive model for disaster risk levels using deep learning with climate and geographical variables. The simulated data includes variables such as average temperature, precipitation, humidity, wind speed, elevation, land cover type, and population density. The model was designed to classify disaster risk into four categories: no risk, low risk, moderate risk, and high risk. Data preprocessing involved normalization and an 80:20 train-test split. The model was trained using the Adam optimization algorithm with activation functions suitable for multi-class classification. Evaluation results show that the model can accurately predict disaster risk levels. This study demonstrates that simulated data can effectively support disaster risk prediction when observational data is limited. With further development using more comprehensive data, this model has the potential to be implemented in an early warning system to support decision-making in climate change mitigation in Indonesia.