<![CDATA[Earthquake-induced soil liquefaction is a major hazard for urban areas in Indonesia built on recent alluvial deposits. This study evaluates three supervised machine learning algorithms Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM) Â liquefaction hazard mapping in Samarinda, East Kalimantan. Four predictor variables were used: shear wave velocity (Vs30), peak ground acceleration (PGA), groundwater level (GWL), and slope. Because only four Cone Penetration Test (CPT) locations were available, the dataset was expanded to 532 labeled samples through stratified spatial sampling of a previously published CPT-based fuzzy GIS liquefaction map. An 80:20 train-test split was used, and SMOTE was applied only to the training data to address class imbalance. Model performance was evaluated using accuracy, precision, recall, F1-score, and confusion matrices. DT achieved the best performance with an accuracy and weighted F1-score of 0.87, followed by SVM (0.80) and RF (0.78). The resulting hazard map classified Samarinda into Very Low (68%), Low (6%), High (7%), and Very High (19%) hazard zones. High-risk areas were concentrated in the eastern part of the city, where shallow groundwater, Sulfaquent soils, and recent alluvial deposits are present.]]>
Copyrights © 2026
