MACHINE LEARNING MODELING OF LANDSLIDE SUSCEPTIBILITY USING THE RANDOM FOREST METHOD TO STUDY THE IMPACT OF INTEGRATED ROCK CHARACTERISTIC PARAMETERS (CASE STUDY: LEBAK REGENCY)

• Main Author: Rizaldy Wibowo, Fariz
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/87063
• Institution: Institut Teknologi Bandung

Landslides are natural disasters that can cause significant damage to infrastructure and threaten lives. Lebak Regency, Banten Province, with its complex morphology resulting from tectonic activity and geological history, is one of the areas with a high landslide potential. This study aims to develop a landslide susceptibility model using the Random Forest algorithm by integrating rock characteristic parameters to improve prediction accuracy and understanding of landslide causative factors. The rock characteristic parameters used in this study include lithology (type and properties of rocks), rock genesis (formation processes), rock chronology (age and weathering level), and rock infiltration (water absorption capacity). Two models were compared: Model 1, which integrates rock characteristic parameters, and Model 2, which excludes these parameters and only incorporates general parameters such as slope gradient, slope aspect, and land cover. The dataset comprises 95 landslide points, 85 non-landslide points, DEMNAS, land cover, geological formations, and fault/lineation data, which were reclassified and segmented. The data was divided into 80% for training and 20% for testing. The evaluation results showed that Model 1 achieved an AUC score of 0.84, outperforming Model 2, which achieved an AUC score of 0.821. Furthermore, Model 1 successfully identified areas with high landslide potential in Cibeber, Panggarangan, and Bayah Districts. These findings demonstrate that integrating rock characteristic parameters enhances prediction accuracy and provides deeper insights into slope stability. This study is expected to serve as a scientific basis for geospatial-based disaster mitigation and spatial planning in landslide-prone areas.