<![CDATA[This study aims to predict flood risk in the Jakarta Province using the K-Nearest Neighbors (KNN) algorithm. The model is designed to classify regions into three risk levels—low, moderate, and high—based on historical features such as water level and annual flood frequency. Data from 2013 to 2024 provided by the Jakarta Regional Disaster Management Agency (BPBD) served as the basis for model training. The normalization process using Z-Score and Min-Max Scaling proved essential in enhancing KNN performance. Cross-validation (5-Fold Cross Validation) revealed that K=3 yielded the best results. The model achieved accuracy rates between 96% and 99%, with precision, recall, and F1-score exceeding 95%. These results demonstrate that KNN can reliably map flood risks based on numerical data trends. Predictions for the years 2026 to 2028 indicate an increased flood risk in several areas, particularly Central, North, and West Jakarta, which fall into the high-risk category. East and South Jakarta are classified as moderate to high risk, while the Thousand Islands region is expected to remain in the low-risk category. This model can be used as a foundation for developing data-driven mitigation strategies and early warning systems]]>
Copyrights © 2025
