All Issue

2026 Vol.42, Issue 4 Preview Page
Data Augmentation and Characteristic Analysis of Rock Weathering Data Using Distance-Correlation-Based Groupwise Oversampling

Distance Correlation 기반 Group-Based Oversampling 기법을 이용한 암석 풍화 데이터의 증폭 및 특성 분석

Hyung-Koo Yoon1*
윤 형구1*
1Member, Prof., Dept. of Disaster Safety Engineering, Daejeon Univ.
1정회원, 대전대학교 재난안전공학과 교수
*Corresponding Author
31 August 2026. pp. 241-252
PDF XML Citation
Abstract

Data imbalance can degrade the performance of artificial-intelligence-based analyses, and various oversampling techniques have been developed to address this issue. In this study, a distance-correlation-based groupwise oversampling method is proposed to address the limitations of the conventional synthetic minority oversampling technique (SMOTE) by considering dependencies among variables. The method first calculates distance correlations among variables and then applies hierarchical clustering to group variables with similar characteristics. Oversampling is subsequently performed independently within each group. The dataset used in this study comprised measurements of P-wave velocity, S-wave velocity, Young’s modulus, Poisson’s ratio, porosity, and crack density obtained during the rock weathering process. The results showed that the generated data generally preserved the main distribution characteristics of the original dataset. Furthermore, the random seed (S) had little influence on the generated data, whereas the number of nearest neighbors (K) significantly affected both the mean difference and standard deviation retention. In particular, K = 5 better preserved the mean, whereas K = 1 better preserved data variability. The findings demonstrate the feasibility of data augmentation considering structural relationships among variables and support the potential application of the proposed method for data augmentation and AI training dataset construction in engineering fields.

Keywords
Data augmentation
Distance correlation
Oversampling
Rock weathering
SMOTE

데이터 불균형 문제는 인공지능 기반 분석에서 예측 성능 저하를 유발하는 주요 원인 중 하나이며, 이를 해결하기 위한 다양한 데이터 증폭 기법이 활용되고 있다. 본 연구에서는 기존 SMOTE 및 ADASYN의 한계를 개선하기 위하여 변수 간 의존성을 고려한 Distance Correlation 기반 Group-Based Oversampling 기법을 제안하였다. 제안 기법은 변수 간 Distance Correlation을 계산한 후 계층적 군집분석을 이용하여 유사한 특성을 갖는 변수들을 그룹화하고, 각 그룹에 대해 독립적으로 Oversampling을 수행하는 방식으로 구성하였다. 연구에는 암석 풍화 과정에서 측정된 압축파 속도, 전단파 속도, 영의 계수, 포아송비, 간극률 및 균열밀도 데이터를 활용하였다. 분석 결과, 생성 데이터는 대부분의 물리량에서 원본 데이터의 주요 분포 특성을 유지하는 것으로 나타났다. 또한 난수 시드(S)의 영향은 제한적인 반면 최근접 이웃 수(K)는 생성 데이터의 평균 차이율과 표준편차 유지율에 상대적으로 큰 영향을 미치는 것으로 확인되었다. 특히 K=5 조건은 평균 특성 보존에, K=1 조건은 분산 특성 보존에 유리한 것으로 나타났다. 본 연구는 변수 간 구조적 관계를 고려한 데이터 증폭 가능성을 제시하였으며, 향후 공학 분야의 데이터 증폭 및 인공지능 학습 데이터 구축에 활용될 수 있을 것으로 기대된다.

References
1

Akosah, S. and Gratchev, I. (2026), “Physics-informed Machine Learning for Imbalance Rockfall Prediction: A Framework Integrating Geotechnical Synthesis and Cost-sensitive Learning”, Geodata and AI, 100055, https://doi.org/10.1016/j.geoai.2026.100055.

10.1016/j.geoai.2026.100055
2

Baek, S. H. and Cho, J. W. (2025), “Field Study on the Correlation between Intelligent Compaction Measurement Values and Location-Specific In-Situ Test Results”, Journal of the Korean Geotechnical Society, Vol.41, No.6, pp.237-250, https://doi.org/10.7843/kgs.2025.41.6.237.

10.7843/kgs.2025.41.6.237
3

Cho, S. E. (2026), “Probabilistic Stability Analysis of River Levees Using Machine-Learning-Based Surrogate Model”, Journal of the Korean Geotechnical Society, Vol.42, No.1, pp.137-150, https://doi.org/10.7843/kgs.2026.42.1.137.

10.7843/kgs.2026.42.1.137
4

Kim, K. S. (2025), “Deep Learning-based Prediction of the Axial Bearing Capacity of PHC Auger-drilled Piles Using Dynamic Load Test Data”, Journal of the Korean Geotechnical Society, Vol.41, No.5, pp.143-155, https://doi.org/10.7843/kgs.2025.41.5.143.

10.7843/kgs.2025.41.5.143
5

Kim, J. (2026), “XGBoost-Based Prediction of Sewer-Induced Sinkholes Under Imbalanced Data Conditions”, Journal of the Korean Geotechnical Society, 42, No.1, pp.81-90, https://doi.org/10.7843/kgs.2026.42.1.81.

10.7843/kgs.2026.42.1.81
6

Li, Y., Chen, J., Fang, Q., Zhang, D., and Huang, W. (2025), “Towards Automated Lithology Classification in NATM Tunnel: A Data-driven Solution for Multi-dimensional Imbalanced Data”, Rock Mechanics and Rock Engineering, Vol.58, No.2, pp.2349-2366, https://doi.org/10.1007/s00603-024-04287-6.

10.1007/s00603-024-04287-6
7

Wang, L., Wang, S., and Pan, Q. (2026), “Data-driven Predictions of Shield Attitudes Using Bayesian Machine Learning Incorporating Cross-correlations and Spatial Correlations”, European Journal of Environmental and Civil Engineering, Vol.30, No.1, pp.1-35, https://doi.org/10.1080/19648189.2025.2564351.

10.1080/19648189.2025.2564351
8

Yoon, H. K. (2025), “Quantitative Assessment of Crack Density Considering Weathering-Induced Uncertainty in Rock Properties”, Journal of the Korean Geotechnical Society, Vol.41, No.6, pp.195-204, https://doi.org/10.7843/kgs.2025.41.6.195.

10.7843/kgs.2025.41.6.195
9

Yoon, H. K., Kim, S. Y., Lee, D., and Yu, J. D. (2022), “A Study on the Characteristics of Applying Oversampling Algorithms to Fosberg Fire-Weather Index (FFWI) Data”, Smart Structures and Systems, pp.9-15, https://doi.org/10.12989/sss.2024.34.1.009.

10.12989/sss.2024.34.1.009
10

Zhang, Q., Zhang, H., and Wang, L. (2025), “Influence of Characteristics of Geotechnical Parameter Random Field on Slope Stability”, Structures, Vol.81, 110445. https://doi.org/10.1016/j.istruc.2025.110445.

10.1016/j.istruc.2025.110445
11

Zhao, Y. and Teng, C. (2025), “Classification of Soil Layers in Deep Cement Mixing Using Optimized Random Forest Integrated with AB-SMOTE for Imbalance Data”, Computers and Geotechnics, Vol.179, 106976, https://doi.org/10.1016/j.compgeo.2024.106976.

10.1016/j.compgeo.2024.106976
Information
  • Publisher :The Korean Geotechnical Society
  • Publisher(Ko) :한국지반공학회
  • Journal Title :Journal of the Korean Geotechnical Society
  • Journal Title(Ko) :한국지반공학회 논문집
  • Volume : 42
  • No :4
  • Pages :241-252
  • Received Date : 2026-08-06
  • Revised Date : 2026-08-13
  • Accepted Date : 2026-08-13
  • DOI :https://doi.org/10.7843/kgs.2026.42.4.241
Journal Informaiton Journal of the Korean Geotechnical Society Journal of the Korean Geotechnical Society
Archives
: Vol.23~28 (2007~2012)
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close