Explainable Prediction of Asphalt Pavement Rutting Depth Using XGBoost and SHAP: Insights from the LTPP Database

Jiaxuan He

doi:10.54097/rj5fns32

Authors

Jiaxuan He

DOI:

https://doi.org/10.54097/rj5fns32

Keywords:

XGBoost, SHAP, Asphalt Pavement Rutting

Abstract

Rutting is one of the most critical distresses affecting the performance and durability of asphalt pavements. This study developed an explainable machine learning framework for predicting asphalt pavement rutting depth by integrating XGBoost and SHAP interpretation methods based on the Long-Term Pavement Performance (LTPP) database. A total of 623 valid records from 357 control asphalt pavement sections in warm climate zones with or without freezing conditions were selected. Six key influencing factors—including Annual Average Daily Traffic (AADT), average temperature, precipitation, freeze index, humidity, and radiation—were used as input variables. The XGBoost regression model achieved excellent predictive performance on the test set, with an R² of 0.9434, RMSE of 0.9278, and MAE of 0.4083. SHAP analysis revealed that AADT and average temperature are the dominant factors contributing 42.1% and 34.6% respectively to rut depth prediction, forming a clear “dual-core driving” mechanism. Precipitation showed moderate influence, while freeze-thaw, humidity, and radiation had relatively minor effects. The results highlight the strong nonlinear coupling between traffic load and high-temperature environment in rut development. This study not only provides a high-precision prediction model but also offers interpretable insights that can support scientific decision-making in anti-rutting pavement design, maintenance strategies, and regional management in high-traffic and high-temperature areas.

Downloads

Download data is not yet available.

References

ALNAQBI A J, ZEIADA W, AL-KHATEEB G G, et al. Creating rutting prediction models through machine learning techniques utilizing the long-term pavement performance database [J]. Sustainability, 2023, 15(18): 13653. DOI: https://doi.org/10.3390/su151813653

[2] AL-SABAEEI A M, SOULIMAN M I, JAGADEESH A. Smartphone applications for pavement condition monitoring: A review [J]. Construction and Building Materials, 2024, 410. DOI: https://doi.org/10.1016/j.conbuildmat.2023.134207

[3] CAO W M, LIU Q F, HE Z Q. Review of Pavement Defect Detection Methods [J]. Ieee Access, 2020, 8: 14531-44. DOI: https://doi.org/10.1109/ACCESS.2020.2966881

[4] ELSEICY A, ALONSO-DíAZ A, SOLLA M, et al. Combined Use of GPR and Other NDTs for Road Pavement Assessment: An Overview [J]. Remote Sensing, 2022, 14(17). DOI: https://doi.org/10.3390/rs14174336

[5] GOLANBARI B, MARDANI A, FARHADI N, et al. Applications of machine learning in predicting rut depth in off-road environments [J]. Scientific Reports, 2025, 15(1). DOI: https://doi.org/10.1038/s41598-025-90054-8

[6] ZHAO Y L, GAO Y, ZHANG K, et al. Establishment of Rutting Model of Wheel-Tracking Test for Real-Time Prediction of Rut Depth of Asphalt Layers [J]. Advances in Civil Engineering, 2021, 2021. DOI: https://doi.org/10.1155/2021/5549891

[7] CHENG C, WANG L, ZHOU X, et al. Predicting rutting development using machine learning methods based on RIOCHTrack data [J]. Applied Sciences, 2024, 14(8): 3177. DOI: https://doi.org/10.3390/app14083177

[8] ZHAO J, WANG H. Machine learning based pavement performance prediction for data-driven decision of asphalt pavement overlay [J]. Structure and infrastructure engineering, 2025, 21(6): 940-55. DOI: https://doi.org/10.1080/15732479.2023.2258498

[9] CHENG T J, YAO C F, DUAN J N, et al. Prediction of active length of pipes and tunnels under normal faulting with XGBoost integrating a complexity-performance balanced optimization approach [J]. Computers and Geotechnics, 2025, 179. DOI: https://doi.org/10.1016/j.compgeo.2024.107048

[10] MEHDARY A, CHEHRI A, JAKIMI A, et al. Hyperparameter Optimization with Genetic Algorithms and XGBoost: A Step Forward in Smart Grid Fraud Detection [J]. Sensors, 2024, 24(4). DOI: https://doi.org/10.3390/s24041230

[11] TAKEFUJI Y. Beyond XGBoost and SHAP: Unveiling true feature importance [J]. Journal of Hazardous Materials, 2025, 488. DOI: https://doi.org/10.1016/j.jhazmat.2025.137382

[12] CHEN S, CAO J, WAN Y, et al. Enhancing rutting depth prediction in asphalt pavements: A synergistic approach of extreme gradient boosting and snake optimization [J]. Construction and Building Materials, 2024, 421: 135726. DOI: https://doi.org/10.1016/j.conbuildmat.2024.135726

[13] GUO R, FU D, SOLLAZZO G. An ensemble learning model for asphalt pavement performance prediction based on gradient boosting decision tree [J]. International Journal of Pavement Engineering, 2022, 23(10): 3633-46. DOI: https://doi.org/10.1080/10298436.2021.1910825

[14] LIU B, JAVED D, HU J, et al. A unified framework for asphalt pavement distress evaluations based on an extreme gradient boosting approach [J]. Coatings, 2025, 15(3): 349. DOI: https://doi.org/10.3390/coatings15030349

[15] ERFANI A, SHAYESTEH N, ADNAN T. Data-augmented explainable AI for pavement roughness prediction [J]. Automation in Construction, 2025, 176: 106307. DOI: https://doi.org/10.1016/j.autcon.2025.106307

[16] AL-JARAZI R, RAHMAN A, AI C F, et al. Development of prediction models for interlayer shear strength in asphalt pavement using machine learning and SHAP techniques [J]. Road Materials and Pavement Design, 2024, 25(8): 1720-38. DOI: https://doi.org/10.1080/14680629.2023.2276412

[17] ERTEN K M, GüRFIDAN R. Regression-Based Performance Prediction in Asphalt Mixture Design and Input Analysis with SHAP [J]. Applied Sciences-Basel, 2025, 15(19). DOI: https://doi.org/10.3390/app151910779

[18] LEI B, CHEN J W, YU Y, et al. An interpretable model for predicting the performances of asphalt mixtures comprising the chemistry composition of steel slag [J]. International Journal of Pavement Engineering, 2025, 26(1). DOI: https://doi.org/10.1080/10298436.2025.2564342

[19] SANDAMAL K, SHASHIPRABHA S, MUTTIL N, et al. Pavement roughness prediction using explainable and supervised machine learning technique for long-term performance [J]. Sustainability, 2023, 15(12): 9617. DOI: https://doi.org/10.3390/su15129617

[20] ADNAN T, ERFANI A. Explainable AI for Predicting Pavement Roughness under Maintenance and No-Maintenance Scenarios [J]. Results in Engineering, 2025: 108666. DOI: https://doi.org/10.1016/j.rineng.2025.108666

[21] ELKINS G E, OSTROM B. Long-term pavement performance information management system user guide [J]. 2021.