<![CDATA[Abstrak Bencana gempa bumi merupakan ancaman nyata di Indonesia, terutama pada daerah yang berada pada zona seismik aktif seperti Kabupaten Madiun. Penelitian ini memiliki tujuan untuk mengembangkan sistem rekomendasi titik evakuasi aman berbasis algoritma K-Nearest Neighbor (KNN) berdasarkan data intensitas gempa dari ShakeMap USGS dan distribusi titik evakuasi dari BPBD. Kriteria yang ditetapkan adalah wilayah dengan nilai Modified Mercalli Intensity (MMI) < 4 dan jarak maksimum 1.000 meter sesuai dengan pedoman BNPB. Pengujian dilakukan terhadap 1.000 titik pengguna simulasi yang tersebar di seluruh wilayah Kabupaten Madiun. KNN berhasil menunjukkan titik evakuasi rekomendasi yang valid sebanyak 370 titik (37.0%), 502 titik (50.2%) ditolak karena melebihi jarak, serta 128 titik (12.8%) gagal karena tidak memiliki titik aman. Temuan ini menunjukkan bahwa KNN efektif untuk mendeteksi awal titik evakuasi, namun keberhasilan KNN bergantung pada distribusi spasial titik evakuasi dan kontur guncangan. Dengan demikian, sistem ini dapat menjadi dasar dalam pengembangan sistem evakuasi berbasis spasial yang responsif dan akurat. Abstract Earthquakes are a real threat in Indonesia, especially in areas located in active seismic zones such as Madiun Regency. This study aims to develop a safe evacuation point recommendation system based on the K-Nearest Neighbour (KNN) algorithm using earthquake intensity data from the USGS ShakeMap and evacuation point distribution data from the Regional Disaster Management Agency (BPBD). The criteria set are areas with a Modified Mercalli Intensity (MMI) value < 4 and a maximum distance of 1,000 metres, in accordance with BNPB guidelines. The testing was conducted on 1,000 simulation user points spread across the entire Madiun Regency area. KNN successfully identified 370 valid recommended evacuation points (37.0%), 502 points (50.2%) were rejected due to exceeding the distance limit, and 128 points (12.8%) failed because they did not have safe points. These findings indicate that KNN is effective for detecting initial evacuation points; however, the success of KNN depends on the spatial distribution of evacuation points and earthquake contours. Thus, this system can serve as a foundation for developing a responsive and accurate spatial-based evacuation system.]]>
Copyrights © 2026
