nav emailalert searchbtn searchbox tablepage yinyongbenwen piczone journalimg journalInfo journalinfonormal searchdiv searchzone qikanlogo popupnotification paper paperNew
基于干散货货主潜类别分析的堆存期选择预测
基金项目(Foundation): 国家自然科学基金青年科学基金(72404044); 中央高校基本科研业务费专项资金资助(3132025293)
邮箱(Email):
DOI: 10.19961/j.cnki.1672-4747.2025.11.003
发布时间: 2026-01-30
出版时间: 2026-01-30
网络发布时间: 2026-01-30
移动端阅读
摘要:

【背景】在当前全球港口加速推进数字化转型和智慧物流体系建设的行业变革浪潮中,港口运营模式正经历着从传统劳动密集型向数据驱动型的根本性转变。【目标】为提升干散货货主堆存期选择的预测精度,解决传统方法难以捕捉货主异质性的问题。【方法】提出一种基于分层融合框架的预测方法——LCA-XGBoost模型,利用潜类别分析(Latent Class Analysis, LCA)识别货主群体的的行为异质性,将其划分为具有不同堆存行为偏好的潜在类别,再将LCA生成的货主堆存行为偏好类型标签作为一项关键的特征增强信息,输入至极限梯度提升算法(eXtreme Gradient Boosting, XGBoost)中,实现对各类货主堆存期选择的精准预测。【数据】采用青岛港前港公司物流电子报文数据及来自上海钢联电子商务的市场相关数据作为基础数据集,包含货主堆存行为与干散货市场动态信息。【结论】LCA揭示出三种典型的货主堆存行为模式,即“均衡稳健型”、“灵活多变型”与“高效敏捷型”,同时LCA-XGBoost模型在准确率、召回率、F1-score及AUC等关键指标上均表现出明显优势,预测精度高,模型表现稳定。【应用】该研究成果可应用于港口智能堆场管理系统,通过精准预测不同类别货主的堆存周期,实现堆场资源的动态优化配置,提升堆场周转率。

Abstract:

[Background] In the current wave of industry transformation in which global ports are accelerating digital transformation and the construction of smart logistics systems, port operations are undergoing a fundamental transformation from labor-intensive to data-driven. [Objective] In order to improve the prediction accuracy of dry bulk cargo owners' storage period selection and address the problem that traditional methods are difficult to capture the heterogeneity of cargo owners. [Method] A prediction method based on a hierarchical fusion framework - the LCA-XGBoost model is proposed. This model utilizes Latent Class Analysis (LCA) to identify the behavioral heterogeneity of the consignor groups, dividing them into potential categories with different preferences for storage behavior. Then, the type labels of the consignor's storage behavior preferences generated by LCA are taken as a key feature-enhancing information and input into the Extreme Gradient Boosting algorithm (XGBoost) to achieve precise prediction of the storage period selection for each type of consignor. [Data] The logistics electronic message data of Qingdao Port Qiangang Company and the market-related data from Shanghai Ganglian e-commerce are used as the basic data set, including the stockpiling behavior of cargo owners and the dynamic information of the dry bulk market. [Conclusion] LCA reveals three typical behavioral patterns of cargo owners, namely the balanced and prudent type, the flexible and adaptable type, and the efficient and agile type. Meanwhile, the LCA-XGBoost model demonstrates distinct advantages in key metrics such as accuracy, recall, F1-score and AUC, featuring high prediction precision and stable performance. [Application] The research results can be applied to the port intelligent storage yard management system. By accurately predicting the storage cycle of different types of cargo owners, the dynamic optimal allocation of storage yard resources can be realized, and the turnover rate of the storage yard can be improved.

参考文献

[1] KAMAKURA W A, RUSSELL G J. A probabilistic choice model for market segmentation and elasticity structure[J]. Journal of Marketing Research, 1989, 26(4): 379-390.

[2] HESS S, TRAIN K. Correlation and scale in mixed logit models[J]. Journal of Choice Modelling, 2017, 23: 1-8..

[3] GREENE W H, HENSHER D A. A latent class model for discrete choice analysis: Contrasts with mixed logit[J]. Transportation Research Part B: Methodological, 2003, 37(8): 681-698.

[4] 潘晓锋, 左志. 基于潜在类别模型的出行决策机理异质性研究[J]. 交通运输工程与信息学报, 2022, 20(1): 55-62.

[5] KIM S H. How heterogeneity has been examined in transportation safety analysis: a review of latent class modeling applications[J]. Analytic Methods in Accident Research, 2023, 40: 100292.

[6] ETZIONI S, DAZIANO R A, BEN-ELIA E, et al. Preferences for shared automated vehicles: a hybrid latent class modeling approach[J]. Transportation Research Part C: Emerging Technologies, 2021, 125: 103013.

[7] 周伊冰, 霍娅敏, 张奕源. 基于潜在类别分析的城市轨道交通通勤市场细分研究[J]. 综合运输, 2021, 43(9): 9-16.

[8] 于二泽, 周继彪. 共享单车用户使用行为与服务感知的影响因素分析[J]. 武汉理工大学学报(交通科学与工程版), 2022, 46(2): 206-212.

[9] 苏焕银, 陶文聪, 彭舒婷, 等. 基于潜在类别模型的城际高速铁路客运市场细分[J]. 铁道科学与工程学报, 2021, 18(6): 1410-1417.

[10] ROMáN C, ARENCIBIA A I, FEO-VALERO M. A latent class model with attribute cut-offs to analyze modal choice for freight transport[J]. Transportation Research Part A: Policy and Practice, 2017, 102: 212-227.

[11] LIAN L, ZHANG S, WANG Z, et al. Customers’ mode choice behaviors of express service based on latent class analysis and logit model[J]. Mathematical Problems in Engineering, 2015, 2015: 610673.

[12] 刘建荣, 郝小妮. 考虑出行者异质性的绿色出行行为研究[J]. 华南理工大学学报(自然科学版), 2019, 47(7): 99-104, 120.

[13] 乔剑锋, 王亚楠, 吕淑然, 等. 基于K-means和LCA的自动驾驶交通事故聚类分析[J]. 中国安全科学学报, 2025, 35(7): 192-200.

[14] VAN CRANENBURGH S, WANG S, VIJ A, et al. Choice modelling in the age of machine learning - Discussion paper[J]. Journal of Choice Modelling, 2022, 42: 100340.

[15] 苏跃江, 温惠英, 袁敏贤, 等. 基于集成学习和居民属性数据的出行方式预测模型[J]. 交通运输系统工程与信息, 2023, 23(3): 153-160.

[16] 李文权, 邓安鑫, 郑炎, 等. 基于机器学习的中型城市居民出行方式选择行为研究[J]. 交通运输系统工程与信息, 2024, 24(2): 13-23.

[17] BEI H, CHEN H, LI L, et al. Joint prediction of travel mode choice and purpose from travel surveys: a multitask deep learning approach[J]. Travel Behaviour and Society, 2023, 33: 100625.

[18] SCHIMOHR K, DOEBLER P, SCHEINER J. Prediction of bike-sharing trip counts: comparing parametric spatial regression models to a geographically weighted XGBoost algorithm[J]. Geographical Analysis, 2023, 55(4): 651-684.

[19] LAHOZ L T, PEREIRA F C, SFEIR G, et al. Attitudes and latent class choice models using machine learning[J]. Journal of Choice Modelling, 2023, 49: 100452.

[20] 李晓东, 曹克让, 匡海波. 基于超参数优化集成学习的出行方式选择研究[J]. 交通运输系统工程与信息, 2024, 24(6): 159-168.

[21] 杨大志, 田雨晗, 于晓桦. 基于XGBoost算法的大学生旅游出行方式预测分析[J]. 交通科技与经济, 2025, 27(1): 36-42.

[22] LI X, SHI L, SHI Y, et al. Exploring interactive and nonlinear effects of key factors on intercity travel mode choice using XGBoost[J]. Applied Geography, 2024, 166: 103264.

[23] 郭姝娟, 耿晓洁, 李纲, 等. 基于机器学习的中欧集装箱货运方式选择预测[J]. 科学技术与工程, 2025, 25(1): 357-364.

[24] CHEN T, GUESTRIN C. XGBoost: a scalable tree boosting system[C]//Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. San Francisco California USA. ACM, 2016: 785-794.

[25] CHAWLA N V, BOWYER K W, HALL L O, et al. SMOTE: synthetic minority over-sampling technique[J]. Journal of Artificial Intelligence Research, 2002, 16: 321-357.

基本信息:

DOI:10.19961/j.cnki.1672-4747.2025.11.003

中图分类号:U695.2

引用信息:

[1]栾建霖,杨彬,王斯妮,等.基于干散货货主潜类别分析的堆存期选择预测[J].交通运输工程与信息学报().DOI:10.19961/j.cnki.1672-4747.2025.11.003.

基金信息:

国家自然科学基金青年科学基金(72404044); 中央高校基本科研业务费专项资金资助(3132025293)

发布时间:

2026-01-30

出版时间:

2026-01-30

网络发布时间:

2026-01-30

检 索 高级检索

引用

GB/T 7714-2015 格式引文
MLA格式引文
APA格式引文