All Issue

2026 Vol.44, Issue 5 Preview Page

Article

Interpretable Analysis of Human Driver Decision-Making using XGBoost-SHAP in Merging Scenarios

XGBoost-SHAP을 활용한 합류 상황에서 인간 운전자 의사결정 요인의 해석적 분석

Seunghyun LEE1
,
Juhui GIM2*
이 승현1
,
김 주희2*
1M.S. Student, School of Mechatronics Engineering, Changwon National University, Changwon 51140, Korea
2Professor, School of Mechatronics Engineering, Changwon National University, Changwon 51140, Korea
1국립창원대학교 메카트로닉스공학부 석사과정
2국립창원대학교 메카트로닉스공학부 교수
*Corresponding Author
31 October 2026. pp. 851-870
PDF XML Citation
Abstract

This paper presents an interpretable framework for analyzing human driver decision-making in merging scenarios. Highway merging represents a complex interactive driving situation where drivers operate under spatial and temporal constraints, yet the underlying decision-making process remains difficult to quantify. To address this issue, the future heading angle is adopted as a proxy for early maneuver responses associated with decision-making. An XGBoost-based model is developed to predict the heading angle using spatial, temporal, and dynamic variables derived from vehicle trajectory data, and SHAP is applied to interpret the contributions of individual variables and their interactions. The analysis results show that driver decision-making is characterized by three key mechanisms: comparative evaluation between current and target lanes, threshold-based behavioral transitions, and dynamic modulation of influencing factors depending on the merging stage. These results indicate that driver behavior follows a condition-dependent decision-making structure shaped by interaction context.

Keywords
driver decision-making
explainable AI
merging scenarios
vehicle interaction
XGBoost-SHAP

본 연구는 합류 상황에서 인간 운전자의 의사결정을 해석하기 위한 설명가능한 분석 프레임워크를 제안한다. 고속도로 합류는 공간적·시간적 제약 하에서 다수 차량 간 상호작용이 발생하는 대표적인 주행 상황이며, 해당 상황에서의 운전자의 의사결정 과정은 복잡하게 형성되며, 이를 정량적으로 해석하는 데에는 한계가 존재한다. 이를 위해 미래 시점의 헤딩각을 의사결정과 연관된 초기 횡방향 기동 반응을 나타내는 대리 변수로 설정한다. 차량 궤적 데이터로부터 도출된 공간적, 시간적, 동적 변수를 활용하여 XGBoost 기반의 예측 모델을 구축하고, SHAP을 적용하여 각 변수의 기여도와 변수 간 상호작용 효과를 해석한다. 분석 결과, 운전자의 의사결정은 세 가지 주요 특성으로 구조화된다. 첫째, 현재 차로와 목표 차로 간 주행 여건을 비교하여 상대적으로 유리한 방향으로 의사결정이 이루어지는 경향을 보인다. 둘째, 변수 값이 특정 수준에 도달할 때 의사결정이 급격히 전환되는 임계적 특성이 나타난다. 셋째, 합류 시점에 따라 주변 요인의 영향력이 달라지는 동적 조정 구조가 확인된다. 이러한 결과는 인간 운전자의 합류 의사결정과 연관된 초기 횡방향 기동 반응이 주변 차량과의 상호작용 맥락에 따라 다르게 나타날 수 있음을 시사한다. 본 연구에서 제안하는 프레임워크는 복잡한 주행 환경에서 운전자 행동을 해석하기 위한 체계적인 접근 방법을 제공하며, 향후 인간 중심의 자율주행 의사결정 모델 개발에 기여할 수 있을 것으로 기대된다.

키워드
운전자 의사결정
설명가능한 인공지능
합류 시나리오
차량 간 상호작용
XGBoost-SHAP
References
1

Adadi A., Berrada M. (2018), Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI), IEEE Access, 6, IEEE, 52138-52160.

10.1109/ACCESS.2018.2870052
2

Bauder M., Paula D., Pfeilschifter C., Petermeier F., Kubjatko T., Riener A. et al. (2024), Influences of Vehicle Communication on Human Driving Reactions: A Simulator Study on Reaction Times and Behavior for Forensic Accident Analysis, Sensors, 24(14), MDPI, 4481.

10.3390/s2414448139065878PMC11281119
3

Chen D., Hajidavalloo M. R., Li Z., Chen K., Wang Y., Jiang L. et al. (2023), Deep Multi-Agent Reinforcement Learning for Highway On-Ramp Merging in Mixed Traffic, IEEE Trans. Intell. Transp. Syst., 24(11), IEEE, 11623-11638.

10.1109/TITS.2023.3285442
4

Chen T., Guestrin C. (2016), XGBoost: A Scalable Tree Boosting System, Proc. 22nd ACM SIGKDD Int. Conf. Knowl. Discov. Data Min., ACM, 785-794.

10.1145/2939672.2939785
5

Gipps P. G. (1986), A Model for the Structure of Lane-Changing Decisions, Transp. Res. B Methodol., 20(5), Elsevier, 403-414.

10.1016/0191-2615(86)90012-3
6

Green M. (2000), How Long Does It Take to Stop? Methodological Analysis of Driver Perception-Brake Times, Transp. Hum. Factors, 2(3), Lawrence Erlbaum Associates, 195-216.

10.1207/STHF0203_1
7

Guidotti R., Monreale A., Ruggieri S., Turini F., Giannotti F., Pedreschi D. (2018), A Survey of Methods for Explaining Black Box Models, ACM Comput. Surv., 51(5), ACM, 93:1-93:42.

10.1145/3236009
8

Gupta A., Choudhary P., Parida M. (2025), Exploring Driver Decision-Making in Lane-Changing: A Human Factors Approach Using AI and Naturalistic Data, Transp. Res. F Traffic Psychol. Behav., 114, Elsevier, 794-820.

10.1016/j.trf.2025.06.030
9

Hu J., He L., Wang R. (2020), Safety Evaluation of Freeway Interchange Merging Areas Based on Driver Workload Theory, Sci. Prog., 103(3), SAGE Publications, 0036850420940878.

10.1177/003685042094087832660349PMC10451929
10

Hugemann W. (2002), Driver Reaction Times in Road Traffic, Proc. 11th EVU Annu. Conf., Portoroz, Slovenia.

11

Kesting A., Treiber M., Helbing D. (2007), General Lane-Changing Model MOBIL for Car-Following Models, Transp. Res. Rec., 1999(1), Transportation Research Board, 86-94.

10.3141/1999-10
12

Kherroubi Z. E. A., Aknine S. (2024), Review of Driver Behaviour Modelling for Highway On-Ramp Merging, IET Intell. Transp. Syst., 18(S1), Wiley, 2793-2813.

10.1049/itr2.12572
13

Li M., Wang Y., Sun H., Cui Z., Huang Y., Chen H. (2023), Explaining a Machine-Learning Lane Change Model With Maximum Entropy Shapley Values, IEEE Trans. Intell. Veh., 8(6), IEEE, 3620-3628.

10.1109/TIV.2023.3266196
14

Li Y., Yuan S., Chang Y., Chen X., Yang Q., Yang Z. et al. (2025), Human-Aligned Safe Reinforcement Learning for Highway On-Ramp Merging in Dense Traffic, arXiv preprint, arXiv:2503.02624.

10.2139/ssrn.5212417
15

Liu H., Chen S., Ma Y., Qiao F., Pang Q., Zhang Z. et al. (2025), High Fuel Consumption Driving Behavior Causal Analysis Based on LightGBM and SHAP, Iran. J. Sci. Technol. Trans. Civ. Eng., 49(2), Springer Nature, 2065-2077.

10.1007/s40996-024-01541-3
16

Liu L., Li X., Li Y., Li J., Liu Z. (2024), Reinforcement-Learning-Based Multilane Cooperative Control for On-Ramp Merging in Mixed-Autonomy Traffic, IEEE Internet Things J., 11(24), IEEE, 39809-39819.

10.1109/JIOT.2024.3447039
17

Lu H., Stepan G., Lu J., Takacs D. (2022), Dynamics of Vehicle Stability Control Subjected to Feedback Delay, Eur. J. Mech. A Solids, 96, Elsevier, 104678.

10.1016/j.euromechsol.2022.104678
18

Lundberg S. M., Lee S.-I. (2017), A Unified Approach to Interpreting Model Predictions, Adv. Neural Inf. Process. Syst., 30, Curran Associates, Inc., 4765-4774.

19

National Highway Traffic Safety Administration (2009), Analysis of Lane-Change Crashes and Near-Crashes, 3-4.

20

Peng Z., Zuo J., Ji H., RengTeng Y., Wang Y. (2024), A Comparative Analysis of Risk Factors in Taxi-Related Crashes Using XGBoost and SHAP, Int. J. Inj. Control Saf. Promot., 31(3), Taylor & Francis, 508-520.

10.1080/17457300.2024.2349555
21

Ran B., Leight S., Chang B. (1999), A Microscopic Simulation Model for Merging Control on a Dedicated-Lane Automated Highway System, Transp. Res. C Emerg. Technol., 7(6), Elsevier, 369-388.

10.1016/S0968-090X(99)00028-5
22

Sekadakis M., Garefalakis T., Moertl P., Yannis G. (2026), Analyzing SHAP Values of XGBoost Algorithms to Understand Driving Features Affecting Take-Over Time From Vehicle Alert to Driver Action, Displays, 91, Elsevier, 103263.

10.1016/j.displa.2025.103263
23

Sun H., Cheng Q., Wang P., Huang Y., Liu Z. (2026), Lane Change Decision Prediction: An Efficient BO-XGB Modelling Approach With SHAP Analysis, Transportmetr. A Transp. Sci., 22(1), Taylor & Francis, 2372020.

10.1080/23249935.2024.2372020
24

Sun J., Li Z., Sun J. (2015), Study on Traffic Characteristics for a Typical Expressway On-Ramp Bottleneck Considering Various Merging Behaviors, Physica A, 440, Elsevier, 57-67.

10.1016/j.physa.2015.08.007
25

Sun J., Yang H. (2024), Learning Two-Dimensional Merging Behaviour From Vehicle Trajectories With Imitation Learning, Transp. Res. C Emerg. Technol., 160, Elsevier, 104530.

10.1016/j.trc.2024.104530
26

Wang H., Liu H., Wang W., Sun L. (2024), On Trustworthy Decision-Making Process of Human Drivers From the View of Perceptual Uncertainty Reduction, IEEE Trans. Intell. Transp. Syst., 25(2), IEEE, 1625-1636.

10.1109/TITS.2023.3316455
27

Xing Y., Wu Y., Wang H., Wang L., Li L., Peng Y. (2025), Failed Lane-Changing Detection and Prediction Using Naturalistic Vehicle Trajectories, Transp. Res. C Emerg. Technol., 170, Elsevier, 104939.

10.1016/j.trc.2024.104939
28

Zhan W., Sun L., Wang D., Shi H., Clausse A., Naumann M. et al. (2019), INTERACTION Dataset: An International, Adversarial and Cooperative Motion Dataset in Interactive Driving Scenarios With Semantic Maps, arXiv preprint, arXiv:1910.03088.

29

Zhang X., Wu L., Liu H., Wang Y., Li H., Xu B. (2023), High-Speed Ramp Merging Behavior Decision for Autonomous Vehicles Based on Multiagent Reinforcement Learning, IEEE Internet Things J., 10(24), IEEE, 22664-22672.

10.1109/JIOT.2023.3304890
Information
  • Publisher :Korean Society of Transportation
  • Publisher(Ko) :대한교통학회
  • Journal Title :Journal of Korean Society of Transportation
  • Journal Title(Ko) :대한교통학회지
  • Volume : 44
  • No :5
  • Pages :851-870
  • Received Date : 2026-04-30
  • Revised Date : 2026-05-27
  • Accepted Date : 2026-07-06
  • DOI :https://doi.org/10.7470/jkst.2026.44.5.851
Journal Informaiton Journal of Korean Society of Transportation Journal of Korean Society of Transportation
Archives
: 2015 Vol.34~
  • scopus
  • NRF
  • KOFST
  • crossref cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close