| 标题 |
SHAP-Driven Pavement Maintenance Prediction and Prioritization Framework Using LightGBM
|
| 网址 | |
| DOI |
暂未提供，该求助的时间将会延长，查看原因？
|
| 其它 |
abstract： This study presents an interpretable machine learning framework for prioritizing pavement maintenance and rehabilitation (M&R) through domain-integrated feature selection and Pareto-based decision modeling. A tuned Light Gradient Boosting Machine (LightGBM) classifier was trained to predict discrete M&R categories across a network of 1,166 pavement sections, achieving strong performance with a test accuracy of 95%, macro-averaged F1-score of 0.94, and precision/recall of 0.95. Feature selection was performed using a hybrid approach that combined Gain Ratio, SHAP (SHapley Additive Explanations), and the Cosine Amplitude Method (CAM), consistently identifying Roughness International Roughness Index (IRI) (m/km) and Current Million Standard Axles (MSA) as the most influential predictors. A third variable, Structural Number (SN), was included based on engineering judgment to ensure representation of pavement structural capacity... |
| 求助人 | |
| 下载 |
PDF的下载单位、IP信息已删除
(2025-6-4)
凌凌
cdercder
无极微光
水煮电吹风
欧克
v0id
学术文献互助
Kao
GingerF
DOC_XIONG
唠叨的大门
Nole
Criminology34
OK
DW
Loscipy
谢雷XIELei
铁锤牛马版
七听
dew
皖公网安备34019202002308
