All Issue

2026 Vol.42, Issue 4 Preview Page
Spatial Encoding Strategies for XGBoost-Based Consolidation Settlement Prediction: A Comparative Study at Songdo New City

XGBoost 기반 압밀 침하 예측을 위한 공간 인코딩 전략: 송도 신도시 사례연구

Thea Raksmey1
,
Yucheol Jeong2
,
Chien Davith3
,
Changho Lee4*
테아 락스메이1
,
정 유철2
,
치엔 다비드3
,
이 창호4*
1Graduate Student, Dept. of Architecture and Civil Engineering, Chonnam National Univ.
2Member, Graduate Student, Dept. of Architecture and Civil Engineering, Chonnam National Univ.
3Graduate Student, Dept. of Architecture and Civil Engineering, Chonnam National Univ.
4Member, Prof., Dept. of Civil Engineering, Chonnam National Univ.
1비회원, 전남대학교 건축토목공학과 석사과정
2정회원, 전남대학교 건축토목공학과 석사과정
3비회원, 전남대학교 건축토목공학과 석사과정
4정회원, 전남대학교 토목공학과 교수
*Corresponding Author
31 August 2026. pp. 75-89
PDF XML Citation
Abstract

Accurately predicting consolidation settlement remains challenging for reclaimed land projects because soft clay deposits exhibit strong spatial variability that conventional machine-learning models may not adequately capture. Raw geographic coordinates provide limited information about spatial relationships, yet most geotechnical prediction studies have not systematically examined how coordinate transformation affects predictive performance. This study compared five spatial encoding strategies within an extreme gradient boosting (XGBoost) regression framework trained and tested on 376 borehole records from Songdo New City, South Korea. The encodings included raw coordinates as a baseline, Euclidean distance to neighboring points (EDNP), buffer boundary distances (BUFF), polynomial coordinate features (POLY), and radial basis functions (RBF). Each strategy was evaluated through 10-fold cross-validation with Bayesian hyperparameter optimization. Performance was assessed using predictive accuracy metrics, residual spatial autocorrelation measured by Moran’s I, and feature contributions calculated using Shapley additive explanations. Among all strategies tested, EDNP achieved the strongest performance, reaching a test R2 of 0.776 and a root-mean-square error of 0.091, compared with an R2 of 0.719 for the raw coordinates. Greater encoding complexity did not guarantee better predictive accuracy; POLY, RBF, and BUFF produced comparable performance despite their structural differences. The index properties of the soils controlled the predictions across all encodings, with spatial features playing a complementary role.

Keywords
Consolidation settlement
Euclidean distance
Machine learning
Spatial encodings
Spatial prediction
XGBoost

매립 지역의 연약 점토층은 높은 공간적 변동성을 보여 종종 통상적인 기계 학습 기법의 적용이 어려워 연약 지반 매립 프로젝트에서 정확한 압밀 침하 예측은 지속적인 과제로 남아 있다. 기계 학습의 공간 입력으로 직접 사용되는 지리적 좌표는 제한적인 공간적 의미를 가지지만, 대부분의 지반 침하 예측 연구에서는 좌표 변환이 예측 성능에 미치는 영향을 체계적으로 검토하지 않았다. 본 연구에서는 다섯 가지 공간 인코딩 기법을 분석하고, 대한민국 송도 신도시의 376개 시추공 자료를 이용하여 학습/검증된 XGBoost 회귀 절차를 사용하여 결과를 비교 분석하였다. 인코딩 기법은 기준이 되는 원시 좌표(raw coordinates), Euclidean distance to neighboring points(EDNP), buffer boundary distances(BUFF), polynomial coordinate features(POLY), radial basis functions(RBF)를 적용하였으며, 각 방법은 Bayesian hyperparameter 최적화를 적용한 10-fold 교차 검증을 통해 평가되었다. 각 기법의 성능은 예측 정확도 평가지표, Moran’s I를 통한 잔류 공간 자기상관, SHAP 기반 특징 기여도의 조합으로 평가하였다. 학습된 모든 방법 중에서 원시 좌표의 R2 = 0.719에 비교하여 EDNP가 R2 = 0.776, RMSE = 0.091로 가장 좋은 성능을 보였다. 인코딩의 복잡도가 높다고 해서 예측 정확도가 반드시 향상되는 것은 아니며, POLY, RBF, BUFF는 구조적 차이에도 불구하고 유사한 성능을 보였다. 지반 정수의 특성이 모든 인코딩의 예측을 좌우하며, 공간 특징은 보완적인 역할을 했다.

References
1

Akiba, T., Sano, S., Yanase, T., Ohta, T., and Koyama, M. (2019), Optuna: A Next-generation Hyperparameter Optimization Framework, http://arxiv.org/abs/1907.10902.

10.1145/3292500.3330701
2

Behrens, T., Schmidt, K., Viscarra Rossel, R. A., Gries, P., Scholten, T., and MacMillan, R. A. (2018), “Spatial Modelling with Euclidean Distance Fields and Machine Learning”, European Journal of Soil Science, Vol.69, No.5, pp.757-770, https://doi.org/10.1111/ejss.12687.

10.1111/ejss.12687
3

Bergstra, J., Bardenet, R., Bengio, Y., and Kégl, B. (2011), “Algorithms for hyper-parameter optimization”, Advances in Neural Information Processing Systems, Vol.24, pp.2546-2554, https://proceedings.neurips.cc/paper_files/paper/2011/hash/86e8f7ab32cfd12577bc2619bc635690-Abstract.html.

4

Biau, G. and Devroye, L. (2015), Lectures on the Nearest Neighbor Nethod, Springer, https://doi.org/10.1007/978-3-319-25388-6.

10.1007/978-3-319-25388-6
5

Cawley, G. C. and Talbot, N. L. C. (2010), “On Over-fitting in Model Selection and Subsequent Selection Bias in Performance Evaluation”, Journal of Machine Learning Research (Vol.11).

6

Chen, T. and Guestrin, C. (2016), XGBoost: A Scalable Tree Boosting System, https://doi.org/10.1145/2939672.2939785.

10.1145/2939672.2939785
7

Chen, W., Ding, J., Shi, C., Wang, T., and Connolly, D. P. (2024), “Geotechnical Correlation Field-informed and Data-driven Prediction of Spatially Varying Geotechnical Properties”, Computers and Geotechnics, Vol.171, https://doi.org/10.1016/j.compgeo.2024.106407.

10.1016/j.compgeo.2024.106407
8

Doğan, A. and Akbaş, S. O. (2026), “A Comparative Analysis of Laboratory, In-situ, and Observational Methods for Predicting the Magnitude and Duration of Consolidation Settlement: Northern Marmara Highway Case History”, International Journal of Geo-Engineering, Vol.17, No.1, https://doi.org/10.1186/s40703-026-00269-5.

10.1186/s40703-026-00269-5
9

Gajan, S. and Kantor, C. (2025), “Modeling of Settlement of Shallow-founded Rocking Structures Using Explainable Physics-guided Machine Learning”, Machine Learning with Applications, Vol.21, 100702, https://doi.org/10.1016/j.mlwa.2025.100702.

10.1016/j.mlwa.2025.100702
10

Gedamu, W. T., Plank-Wiedenbeck, U., and Wodajo, B. T. (2024), “A Spatial Autocorrelation Analysis of Road Traffic Crash by Severity Using Moran’s I Spatial Statistics: A Comparative Study of Addis Ababa and Berlin Cities”, Accident Analysis & Prevention, Vol.200, 107535, https://doi.org/10.1016/j.aap.2024.107535.

10.1016/j.aap.2024.107535
11

Hengl, T., Nussbaum, M., Wright, M. N., Heuvelink, G. B. M., and Gräler, B. (2018), “Random Forest as a Generic Framework for Predictive Modeling of Spatial and Spatio-temporal Variables”, PeerJ, Vol.2018, No.8, https://doi.org/10.7287/peerj.preprints.26693v3.

10.7287/peerj.preprints.26693v3
12

Jalali, U. H., Alshameri, B., Khalid, M. H., Hassan, W., Abdeldjouad, L., Jamil, S. M., Farooq, H., and Maqsood, Z. (2025), “Data-driven Approach to Enhance Deep Foundation Safety: Reliable Methods for Predicting Bored Pile Capacity”, International Journal of Geo-Engineering, Vol.16, No.1, https://doi.org/10.1186/s40703-025-00247-3.

10.1186/s40703-025-00247-3
13

Kim, D., Kim, K. S., Ko, S., Choi, Y., and Lee, W. (2012), “Assessment of Geotechnical Variability of Songdo Silty Clay”, Engineering Geology, 133-134, 1-8, https://doi.org/10.1016/j.enggeo.2012.02.009.

10.1016/j.enggeo.2012.02.009
14

Kim, H. J., Mawuntu, K. B. A., Park, T. W., Kim, H. S., Park, J. Y., and Jeong, Y. S. (2023), “Spatial Autocorrelation Incorporated Machine Learning Model for Geotechnical Subsurface Modeling”, Applied Sciences (Switzerland), Vol.13, No.7, https://doi.org/10.3390/app13074497.

10.3390/app13074497
15

Liu, X., Kounadi, O., and Zurita-Milla, R. (2022), “Incorporating Spatial Autocorrelation in Machine Learning Models Using Spatial Lag and Eigenvector Spatial Filtering Features”, International Journal of Geo-Information, https://doi.org/10.3390/ijgi11040242.

10.3390/ijgi11040242
16

Liu, Y. and Pu, H. (2018), “Settlement Prediction of Ground Treatment with Prefabricated Vertical Drains based on Measured Field Data”, Marine Georesources & Geotechnology, Vol.36, No.4, pp.494-502, https://doi.org/10.1080/1064119X.2017.1340374.

10.1080/1064119X.2017.1340374
17

Lundberg, S. and Lee, S.-I. (2017), A Unified Approach to Interpreting Model Predictions, http://arxiv.org/abs/1705.07874.

18

Ma, T., Ni, H., Luo, X., Zhang, M., Shen, L., Chen, C., Luo, K., Xie, S., Yang, Z., Lin, Y., Lin, H., Peng, K., Fu, X., and Jiang, Z. (2026), “Application of Swarm Intelligence Optimization Techniques and Explainable Data-driven Methods in Predicting the Shear Strength of Serrated Jointed Rock Masses”, International Journal of Geo-Engineering, Vol.17, No.1, https://doi.org/10.1186/s40703-026-00264-w.

10.1186/s40703-026-00264-w
19

Meyer, H. and Pebesma, E. (2022), “Machine Learning-based Global Maps of Ecological Variables and the Challenge of Assessing Them”, Nature Communications, Vol.13, 2208, https://doi.org/10.1038/s41467-022-29838-9.

10.1038/s41467-022-29838-935459230PMC9033849
20

Ohmer, M., Doll, F., and Liesch, T. (2025), “Incorporating Spatial Information for Regionalization of Environmental Parameters in Machine Learning Nodels”, Mathematical Geosciences, Vol.57, No.2, pp.251-273, https://doi.org/10.1007/s11004-024-10163-4.

10.1007/s11004-024-10163-4
21

Osborne, J. W. (2010), “Improving Your Data Transformations: Applying the Box-Cox Transformation”, Practical Assessment, Research & Evaluation, Vol.15, No.12, pp.1-9.

22

Rahimi, A. and Recht, B. (2007), “Random Features for Large-Scale Kernel Machines”, Advances in Neural Information Processing Systems, Vol.20, https://proceedings.neurips.cc/paper_files/paper/2007/file/013a006f03dbc5392effeb8f18fda755-Paper.pdf.

23

Sekulić, A., Kilibarda, M., Heuvelink, G. B. M., Nikolić, M., and Bajat, B. (2020), “Random Forest Spatial Interpolation”, Remote Sensing, 12(10), https://doi.org/10.3390/rs12101687.

10.3390/rs12101687
24

Shahin, M. (2024), “Progression of Artificial Intelligence/machine Learning in Geotechnical Engineering”, Machine Learning and Data Science in Geotechnics, Vol.1, No.1, pp.1-5, https://doi.org/10.1108/MLAG-10-2024-0010.

10.1108/MLAG-10-2024-0010
25

Shan, H., Ai, L., He, C., and Li, K. (2025), “Enhancing Multi-objective Prediction of Settlement around Foundation Pit Using Explainable Machine Learning”, Journal of Civil Structural Health Monitoring, Vol.15, No.7, pp.3113-3134, https://doi.org/10.1007/s13349-025-00985-z.

10.1007/s13349-025-00985-z
26

Soleimanfar, M. R., Shirinabadi, R., Hosseini Alaee, N., Moosavi, E., and Mohammadi, G. (2025), “Spatial Correlation Effects on Rock Mass Behavior: Insights from Stochastic Modeling in Longwall Mining”, International Journal of Geo-Engineering, Vol.16, No.1, https://doi.org/10.1186/s40703-025-00245-5.

10.1186/s40703-025-00245-5
27

Terzaghi, K. (1943). Theoretical Soil Mechanics. John Wiley & Sons.

10.1002/9780470172766
28

Tobler, W. R. (1970), “A Computer Movie Simulating Urban Growth in the Detroit Region”, Economic Geography, Vol.46, pp.234-240, https://doi.org/10.2307/143141.

10.2307/143141
29

Wang, Z. Z., Guo, X., and Hu, Y. (2025), “Characterisation of the Spatial Variability of Underconsolidated Singapore Marine Clay Using Random Field Theory”, Georisk (Vol.19, Number 2, pp.448-466). Taylor and Francis Ltd., https://doi.org/10.1080/17499518.2024.2422487.

10.1080/17499518.2024.2422487
30

Xie, J., Huang, J., Zeng, C., Huang, S., and Burton, G. J. (2022), “A Generic Framework for Geotechnical Subsurface Modeling with Machine Learning”, Journal of Rock Mechanics and Geotechnical Engineering, Vol.14, No.5, pp.1366-1379, https://doi.org/10.1016/j.jrmge.2022.08.001.

10.1016/j.jrmge.2022.08.001
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 :75-89
  • Received Date : 2026-06-12
  • Revised Date : 2026-07-06
  • Accepted Date : 2026-07-07
  • DOI :https://doi.org/10.7843/kgs.2026.42.4.75
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