| 吴梅玉,钟世彬,徐岩.基于机器学习的江西省高速公路低效用地影响因素研究[J].情报工程,2026,(3):114-127 |
| 基于机器学习的江西省高速公路低效用地影响因素研究 |
| Research on the Influencing Factors of Inefficient Land Use on Expressways in Jiangxi Province Based on Machine Learning |
| |
| DOI： |
| 中文关键词: 高速公路低效用地；影响因素；机器学习；驱动机制 |
| 英文关键词: Inefficient Land Use of Expressways; Influencing Factors; Machine Learning; Driving Mechanism |
| 基金项目:江西省自然资源科技创新项目“‘双碳’战略目标下的优化国土空间用途管制研究”（ZRKJ20232406）。 |
| 作者 | 单位 | | 吴梅玉 | 1. 江西省国土空间调查规划研究院 南昌 330025；2. 自然资源部大湖流域国土空间生态保护修复工程技术创新中心 南昌 330025 | | 钟世彬 | 1. 江西省国土空间调查规划研究院 南昌 330025；2. 自然资源部大湖流域国土空间生态保护修复工程技术创新中心 南昌 330026 | | 徐岩 | 1. 江西省国土空间调查规划研究院 南昌 330025；2. 自然资源部大湖流域国土空间生态保护修复工程技术创新中心 南昌 330027 |
|
| 摘要点击次数: 18 |
| 全文下载次数: 14 |
| 中文摘要: |
| [目的/意义]探索高速公路低效用地的影响因素及形成机制，是推进高速公路低效用地再开发利用的关键基础之一。[方法/过程] 研究以江西省高速公路低效用地为研究对象，从自然、社会经济、政策三个维度选取10 个分析指标，构建融合Logistic 回归、随机森林与XGBoost 三种机器学习模型的驱动因素分析方法，一定程度上弥补单一模型的局限性，实现对高速公路低效用地影响因素作用方向及影响程度的全面分析。 [结果/结论] Logistic 回归模型所得ROC 曲线面积超过0.8，模型精确度较高；政策规划、GDP、距铁路距离、距服务区距离与距收费站距离对高速公路用地低效具有负向影响；海拔、高速类别、人口与距地灾点距离则对高速公路用地低效具有正向影响。随机森林模型与XGBoost 模型精确度均为90%，准确率分别为90%、89%，召回率分别为90.73%、90.5%，精确度较高，进行加权平均得到综合权重，GDP、高速类别与人口的特征权重值分别为25%、19%、10%，对高速公路用地低效具有显著作用；政策规划、海拔、距收费站距离与距建成区距离，亦对高速公路低效与否产生重要影响；距铁路、服务区与地灾点距离的权重值均小于5%，其影响较小；多因素综合影响下的交通需求与交通供给不匹配是导致高速公路用地低效的关键。解决交通需求与交通供给两者之间的错位，构建绿色高速公路用地体系，提升高速公路运营网络运行效率与服务水平，对提高高速公路利用效率、促进高速公路可持续发展具有重要意义。 |
| 英文摘要: |
| [Objective/Significance] Exploring the influencing factors and mechanisms of inefficient land use in expressways is one of the key foundations for promoting the re-development and utilization of such land. [Methods/Processes] This study takes the inefficient land use in expressways in Jiangxi Province as the research object, selects 10 analysis indicators from three dimensions of nature, social economy and policy, and builds a driving factor analysis method integrating three models of Logistic regression, random forest and XGBoost in machine learning, which to some extent compensates for the limitations of a single model and achieves a comprehensive analysis of the direction and degree of influence of the factors affecting the inefficient land use in expressways. [Results/Conclusions] The area under the Receiver Operating Characteristic Curve (ROC) of the Logistic regression model exceeds 0.8, indicating a high level of accuracy. Policy planning, GDP, distance from railway, distance from service area and distance from toll station have a negative impact on the inefficient land use in expressways; altitude, expressway category, population, distance from geological disaster points and distance from built-up areas have a positive impact. The accuracy of the random forest model and the XGBoost model is 90%, with accuracy rates of 90% and 89% respectively; the recall rates are 90.73% and 90.5% respectively, indicating a high level of accuracy. The comprehensive weights are obtained by weighted averaging. The feature weights of GDP, expressway category and population are 25%, 19% and 10% respectively,which have a significant impact on the inefficient land use in expressways. Policy planning, altitude, distance from toll station and distance from built-up areas also have an important impact on whether the expressway land use is inefficient or not. The weights of distance from railway, service area and geological disaster points are all less than 5%, indicating a relatively small impact. The mismatch between traffic demand and supply under the combined influence of multiple factors is the key to the inefficient land use in expressways. Solving the misalignment between traffic demand and supply, building a green expressway land use system, and improving the operational efficiency and service level of the expressway network are of great significance for improving the utilization efficiency of expressways and promoting their sustainable development. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
