All Issue

2026 Vol.27, Issue 1
Application of Machine Learning Classification Models Using Steep Slope Data: A Case Study on Disaster Risk Assessment

급경사지 데이터 기반 기계학습 분류모형의 적용: 급경사지 재해위험도 평가 사례 분석

Hongyeol Lee1
,
Donggeun Kim2
,
Yoseok Jeong3
,
Sewook Oh4*
이 홍열1
,
김 동근2
,
정 유석3
,
오 세욱4*
1PhD Program, Dept. of Const. and Disaster Prevention Eng., Kyungpook National University
2Prof. Department of Forest Ecology and Protection, Kyungpook National University
3Associate Prof. Dept. of Const. and Disaster Prevention Eng., Kyungpook National University
4Prof. Dept. of Const. and Disaster Prevention Eng., Kyungpook National University
*Corresponding Author
1 January 2026. pp. 5-14
PDF XML Citation
Abstract

To prevent slope-related disasters, qualitative assessment tools such as the “Landslide Risk Assessment Table” and the “Disaster Risk Evaluation Table” have been developed and widely applied in evaluating the hazard potential of steep slopes. Various methodologies have been proposed to develop and improve these tools, with many studies using the Analytic Hierarchy Process (AHP) based on expert judgment. In this study, a database of slope collapse occurrence and non-occurrence cases was established to enable quantitative prediction of slope collapse events. Using this database, several machine learning classification models—including logistic regression, random forest, extremely randomized trees, and XGBoost—were applied and their predictive performance compared. The evaluation was based on disaster risk assessments conducted for steep slopes in Korea. Among the models tested, XGBoost demonstrated the best classification performance, achieving an accuracy of 98.60%, sensitivity of 76.47%, F1 score of 76.47%, and MCC of 75.75% on the test data. This study contributes to the advancement of risk assessment for steep slopes by integrating quantitative prediction methods into traditionally qualitative evaluation frameworks, thereby providing a scientific foundation for decision-making of disaster prevention.

Keywords
Steep slope
Slope collapse
Disaster Risk Assessment Table
Machine learning

비탈면 재해를 예방하기 위해 ‘산사태 위험지 평가표’ 및 ‘재해위험도 평가표’와 같은 정성적 평가 도구들이 개발되어 급경사지의 위험성 평가에 활용되고 있다. 이러한 평가표의 개발과 개선을 위해 다양한 방법론이 제안되고 있으며, 특히 전문가 그룹을 대상으로 한 계층분석기법(AHP)을 이용한 연구가 다수 수행되어 왔다. 본 연구에서는 비탈면 붕괴 발생 여부에 대한 정량적 예측을 목적으로 비탈면 붕괴 발생 및 미발생 사례에 대한 데이터베이스를 구축하고, 이를 기반으로 다양한 기계학습 기법을 적용하여 분류 예측 성능을 비교·평가하였다. 주요 분석 기법으로는 로지스틱 회귀, 랜덤 포레스트, 극단적 무작위 트리, XGBoost 등을 사용하였으며, 국내 급경사지의 재해위험도 평가 결과를 바탕으로 각 모델의 성능을 시험하였다. 그 결과, XGBoost 모형이 시험용 데이터에서 정확도 98.60%, 민감도 76.47%, F1 score 76.47%, MCC 75.75%를 기록하며, 가장 우수한 분류 성능을 나타냈다. 본 연구는 정성적 평가의 비중이 큰 기존의 위험성 평가 방식에 정량적 예측 기법을 도입함으로써, 급경사지 재해 예방을 위한 과학적 의사결정 기반을 수립하기 위한 기초적인 정보를 제공할 수 있을 것으로 기대된다.

키워드
급경사지
비탈면 붕괴
재해위험도평가
머신러닝
References
1

Goodfellow, I., Bengio, Y. and Courville, A. (2016), Deep Learning, MIT Press.

10.23000/TRKO201600014868.169
2

Kim, S.K. and Kim, S.H. (2024), Analysis of Slope Failure Factors Based on Data and Prediction of Countermeasure Methods Using Artificial Intelligence, The Journal of Engineering Geology, Vol. 34, No. 4, pp. 563~575.

3

Korea Slope Safety Association (2022), 2022 National Survey on Steep-Slope Areas, Ministry of the Interior and Safety.

4

Korea Slope Safety Association (2023), 2023 National Survey on Steep-Slope Areas, Ministry of the Interior and Safety.

5

Korea Slope Safety Association (2024), 2024 National Survey on Steep-Slope Areas, Ministry of the Interior and Safety.

6

Ma, J. and Yun, T. S. (2022), Prediction of Slope Failure Arc Using Multilayer Perceptron, Journal of the Korean Geotechnical Society, Vol. 38, No. 8, pp. 39~52.

7

Nagarajah, T. and Poravi, G. (2019), A review on automated machine learning (AutoML) systems, Proceedings of the 2019 5th International Conference for Convergence in Technology (I2CT), pp. 1~6.

10.1109/I2CT45611.2019.9033810
8

Nam, K.H., Kim, M.I., Fawu, Kwan, O.I., Wang, F. and Jeong, G.C. (2020), Prediction of Landslides and Determination of Its Variable Importance Using AutoML, The Journal of Engineering Geology, Vol. 30, No. 3, pp. 315~325.

9

Nam, K.H. and Wang, F.W. (2019), The performance of using an autoencoder for prediction and susceptibility assessment of landslides: a case study on landslides triggered by the 2018 eHokkaido Eastern Iburi earthquake in Japan, Geoenvironmental Disasters, Vol. 6, No. 19, pp. 1~14.

10.1186/s40677-019-0137-5
Information
  • Publisher :Korean Geo-Environmental Society
  • Publisher(Ko) :한국지반환경공학회
  • Journal Title :Journal of the Korean Geo-Environmental Society
  • Journal Title(Ko) :한국지반환경공학회 논문집
  • Volume : 27
  • No :1
  • Pages :5-14
  • Received Date : 2025-08-19
  • Revised Date : 2025-08-21
  • Accepted Date : 2025-12-17
  • DOI :https://doi.org/10.14481/jkges.2026.27.1.5
Journal Informaiton Journal of the Korean Geo-Environmental Society Journal of the Korean Geo-Environmental Society
Online Submission
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • orcid
Journal Informaiton Journal Informaiton - close