随着全球经济一体化和信息技术的飞速发展，物流行业已成为推动地区经济增长的关键因素之一。陕西省作为中国内陆省份的重要一员，其物流业的发展状况不仅影响着本省的经济发展，也对周边地区产生着深远的影响。因此，对陕西省物流需求进行科学、准确的预测，不仅是优化资源配置、提升物流效率的关键，也是推动区域经济一体化、促进经济高质量发展的必然要求。

本研究基于物流需求国内外研究现状和机器学习预测的内涵与步骤等相关理论，紧密结合陕西省物流行业发展现状，开展了一系列研究。首先，在物流需求预测指标体系的构建方面，通过详细分析陕西省物流需求影响因素，同时遵循指标选取原则，从地区经济水平、产业结构、内外贸易、消费市场规模及消费水平和交通运输基础设施建设等五大方面初步构建了预测指标体系，然后基于1990-2023年陕西省物流需求相关数据，运用灰色关联分析法对初始指标体系进行筛选优化，最终确立了包含1个衡量指标和12个物流需求影响因素指标的陕西省物流需求预测指标体系；其次，在物流需求预测模型的构建方面，本研究构建了XGBoost、CatBoost和DT等单一预测模型，进而提出了Stacking-XGBoost-CatBoost-DT组合预测模型，通过模型训练和参数调整得到各预测模型预测性能评价结果，并对其进行对比分析，结果表明：Stacking组合预测模型的MSE、MAE、MAPE、R²等评价指标的评价结果在四个模型中均呈现出最优效果，具备较强的预测性能，从而验证了组合预测模型在陕西省物流需求预测中的可行性和适用性，随后利用该模型对陕西省物流需求进行了科学预测。

最后，本研究针对陕西省物流业的高质量发展提出了系列建议与对策，主要包括：（1）强化区域经济与物流协同发展机制；（2）构建与产业结构适配的现代化物流体系；（3）完善内外贸一体化的物流服务网络；（4）提升消费驱动的物流服务品质；（5）推进交通基础设施智能化升级。这些举措旨在夯实陕西省物流业发展根基，助力陕西省实现经济跨越发展和辐射带动整个西北地区，为区域经济社会的繁荣注入强劲动力。

With the rapid development of global economic integration and information technology, the logistics industry has become one of the key factors driving regional economic growth. As an important member of the inland provinces of China, the development status of the logistics industry in Shaanxi Province not only affects the economic development of the province itself, but also has a profound impact on the surrounding areas. Therefore, conducting scientific and accurate predictions of the logistics demand in Shaanxi Province is not only the key to optimizing resource allocation and improving logistics efficiency, but also an inevitable requirement for promoting regional economic integration and high-quality economic development.

This study is grounded in the relevant theories of domestic and international research on logistics demand, as well as the connotation and procedures of machine learning prediction. It is closely integrated with the current state of logistics development in Shaanxi Province to conduct a series of investigations. Firstly, in the construction of the logistics demand prediction indicator system, through a detailed analysis of the influencing factors of logistics demand in Shaanxi Province and adherence to the principles of indicator selection, a preliminary prediction indicator system was established from five major aspects: regional economic level, industrial structure, domestic and foreign trade, consumer market size and consumption level, and transportation infrastructure construction. Subsequently, based on logistics demand-related data of Shaanxi Province from 1990 to 2023, the grey relational analysis method was employed to screen and optimize the initial indicator system, ultimately resulting in a logistics demand prediction indicator system for Shaanxi Province that includes one measurement indicator and twelve influencing factors. Secondly, in the construction of logistics demand prediction models, this study developed single prediction models such as XGBoost, CatBoost, and DT, and further proposed a Stacking-XGBoost-CatBoost-DT ensemble prediction model. Through model training and parameter adjustment, the predictive performance evaluation results of each model were obtained and comparatively analyzed. The results demonstrate that the Stacking ensemble prediction model exhibits the best performance across all four models in terms of evaluation metrics such as MSE, MAE, MAPE, and R², indicating strong predictive capability. This validates the feasibility and applicability of the ensemble prediction model for logistics demand forecasting in Shaanxi Province. In addition, the model was utilized to conduct a scientific prediction of logistics demand in Shaanxi Province.

Finally, this study proposes a series of recommendations and countermeasures to promote the high-quality development of the logistics industry in Shaanxi Province, which mainly include:(1) Strengthening the coordinated development mechanism between regional economies and logistics; (2) Establishing a modern logistics system aligned with the industrial structure; (3) Improving the logistics service network for integrated domestic and foreign trade; (4) Enhancing consumption-driven logistics service quality; (5) Advancing the intelligent upgrading of transportation infrastructure. These measures aim to consolidate the foundation for the development of Shaanxi’s logistics industry, facilitate the province’s economic leapfrog development, and enhance its radiating influence across the entire Northwest region, thereby injecting strong momentum into regional economic and social prosperity.

[1]杜海涛,韩鑫,靳博.流动中国释放更多生机活力[N].人民日报,2023-06-09(018).

[2]孙剑青.北京市物流需求预测研究[D].北京交通大学,2016.

[3]中华人民共和国国务院.中共中央国务院关于加快建设全国统一大市场的意见[N].人民日报,2022-04-11(001).

[4]中华人民共和国国务院.“十四五”现代物流发展规划[EB/OL].(2022-05-17)[2023-1-10].https://www.gov.cn/gongbao/content/2023/content_5736713.htm.

[5]中共中央办公厅、国务院办公厅.关于推进以县城为重要载体的城镇化建设的意见[EB/OL].(2022-05-06)[2022-5-20].https://www.gov.cn/gongbao/content/2022/content_5690990.htm.

[6]广东省发展改革委.广东省“十四五”现代流通体系建设实施方案[EB/OL].(2022-12-02)[2022-12-05].http://drc.gd.gov.cn/ywtz/content/post_4057830.html.

[7]Fite JT, Don Taylor G, Usher JS, English JR, Roberts JN. Forecasting freight demand using economi-c.indices [J]. International Journal of Physical Distribution Logistics Management. 2002, 32(04): 299- 308.

[8]Nuzzolo A, Comi A. Urban freight demand forecasting: A mixed quantity delivery vehicle-based mo-del[J]. Transportation Research Part E: Logistics and Transportation Review. 2014, 65: 84-98.

[9]Nguyen TY. Research on Logistics Demand Forecast in Southeast Asia [J]. World Journal of Engine-ering and Technology. 2020, 08 (03): 249-256.

[10]Dhulipala S,Patil G R. Freight production of agricultural commodities in India using multiple linear regression and generalized additive modelling [J]. Transport Policy, 2020, 97: 245-258.

[11]Reza M. The relationship between logistics and economic development in Indonesia: Analysis of ti-me series data [J]. Jurnal Teknik Industri: Jurnal Keilmuan dan Aplikasi Teknik Industri, 2013, 15(02): 119-124.

[12]Liimatainen H, Kallionpää E, Pöllänen M, et al. Decarbonizing road freight in the future—Detailed sce narios of the carbon emissions of Finnish road freight transport in 2030 using a Delphi method approach [J]. Technological Forecasting and Social Change, 2014, 81: 177-191.

[13]李忠民,于庆岩.物流促进经济增长的空间异质性研究—以“新丝绸之路”经济带为例[J].经济问题,2014,08(06):121-125.

[14]邱慧,黄解宇,董亚兰.基于灰色系统模型的山西省物流需求预测分析[J].数学的实践与认识,2016,46 (13):66-70.

[15]李林汉,岳一飞,田卫民.基于PCA与Markov残差灰色模型的京津冀物流能力评价和预测[J].北京交通大学学报(社会科学版),2019,18(02):129-142.

[16]赵文德,刘世明.基于DPSIR-FAM模型的区域物流需求规模预测指标体系研究[J].商场现代化，2020,42(22):30-34.

[17]姜金德,周海花.基于区域经济指标的区域物流需求PCR预测研究—以江苏省为例[J].济南大学学报(社会科学版),2021,31(04):124-132+159-160.

[18]Abolghasemi M, Beh E, Tarr G, et al. Demand forecasting in supply chain: The impact of demand volatility in the presence of promotion [J]. Computers & Industrial Engineering, 2020, 142: 106380.

[19]Munkhdalai L, Park K H, Batbaatar E, et al. Deep learning-based demand forecasting for Korean postal delivery service [J]. IEEE Access, 2020, 8: 188135-188145.

[20]Fuqua D, Hespeler S. Commodity demand forecasting using modulated rank reduction for humanitarian logistics planning [J]. Expert Systems with Applications, 2022, 206: 117753.

[21]Michael W. Babcock,Xiaohua Lu,Jerry Norton. Time series forecasting of quarterly railroad grain carloadings[J]. Transportation Research Part E, 1999, 35(1).

[22]Paulo S.A. Freitas,António J.L. Rodrigues. Model combination in neural-based forecasting[J]. European Journal of Operational Research, 2005, 173(3).

[23]Korpela J,Tuominen M. Inventory forecasting with a multiple criteria decision tool [J]. International journal of production economics, 1996, 45 (1-3): 159-168.

[24]Smyl S. A hybrid method of exponential smoothing and recurrent neural networks for time series forecasting [J]. International Journal of Forecasting. 2020, 36(01): 75-85.

[25]Xue H, Jiang C, Cai B, et al. Research on demand forecasting of retail supply chain emergency logistics based on NRS-GA-SVM[C]//2018 Chinese Control And Decision Conference (CCDC). IEEE, 2018: 3647-3652.

[26]Cong Y, Wang J, Li X. Traffic flow forecasting by a least squares support vector machine with a fruit fly optimization algorithm[J]. Procedia Engineering, 2016, 137: 59-68.

[27]王笛,彭思汗.基于灰色模型的运城市果品冷链物流需求预测研究[J].运城学院学报,2024,42(06):38-44.

[28]许越.黄河流域生鲜农产品冷链物流需求预测研究[J].淮南师范学院学报,2024,26(06):27-35.

[29]孟凡齐,陆芬.基于组合预测的安徽省物流需求预测分析[J].物流科技,2024,47(21):104-108.

[30]张泽明.基于BP神经网络的京津冀物流需求预测[J].中国物流与采购,2024,(19):69-70.

[31]张宇超.基于GM(1,1)模型的内蒙古自治区区域物流需求预测[J].内蒙古统计,2023,(03):32-35.

[32]朱美华,陈长彬,王丽捷.桂林机场航空物流需求预测研究[J].桂林航天工业学院学报,2024,29(06):804-809.

[33]李俊瑜.基于需求分析的福建省生鲜农产品物流发展研究[D].福建农林大学,2018.

[34]古俊杰,刘东.基于BP和RBF神经网络模型的江西省农产品冷链物流需求预测[J].中国储运,2024,(10):62-63.

[35]刘子玲,谢如鹤,廖晶,等.基于灰色回归模型广州市果蔬类生鲜农产品冷链物流需求预测[J].包装工程,2024,45(03):243-250.

[36]尹忠恺,程陈.基于遗传神经网络预测模型的辽宁省物流需求预测[J].辽宁工程技术大学学报(社会科学版),2023,25(05):342-349.

[37]徐晓燕,杨慧敏,吕修凯等.基于山东省不同模型的物流需求预测比较研究 [J].包装工程,2022,43(23):207-215.

[38]窦锦.“一带一路”倡议下甘肃省物流需求预测及实证研究[D].兰州财经大学,2017.

[39]李隽波,孙丽娜.基于多元线性回归分析的冷链物流需求预测[J].安徽农业科学,2011,39 (11):6519-6520+6523.

[40]李俊瑜.基于需求分析的阿拉尔市生鲜农产品物流发展研究[D].福州:福建农林大学,2018.

[41]戎陆庆,黄佩华.基于灰色理论的广西果蔬冷链物流需求及其影响因素预测研究[J].中国农业资源与区划，2017,38(12):227-234.

[42]王新娥,王学剑.新疆城镇居民农产品冷链物流需求预测分析[J].物流技术,2014,33(01):185-186+202.

[43]李夏培.基于灰色线性组合模型的农产品物流需求预测[J].北京交通大学学报(社会科学版),2017,16(01):120-126.

[44]徐妍.基于GM(1,1)模型的陕西省农产品物流需求预测分析[J].辽宁农业科学,2021,(05):41-43.

[45]张志清,杜静.基于二次指数平滑和多元线性回归的宁波市港口物流需求预测分析[J].物流科技,2024,47(17):78-82.

[46]兰洪杰,汝宜红.2008 北京奥运食品冷链物流需求预测分析[J].中国流通经济,2008 (02):19-22.

[47]王新利,赵琨.基于神经网络的农产品物流需求预测研究[J].农业技术经济,2010(02):62-68.

[48]梁毅,徐超飞.基于SVM的区域物流需求建模与预测仿真—以浙江省为例[J].物流研究,2024,(03):54-60.

[49]黄毅,韦志民.基于神经网络的广西物流需求预测模型构建与应用研究[J].沿海企业与科技,2022,(04):24-30.

[50]胡艳娟,胡伟,潘雷霆.一种基于LSTM物流资源需求预测模型[J].长春工业大学学报,2022,43(03):193-201+289.

[51]王娜,李晏丞.基于GA-SVR的江苏省区域物流需求预测研究[J].物流科技,2025,48(05):101-105.

[52]王秀梅.基于权重分配组合法的农产品冷链物流需求趋势预测[J].统计与决策,2018,34 (09):55-58.

[53]李敏杰,王健.基于RBF神经网络的水产品冷链物流需求预测研究[J].中国农业资源与区划,2020,41(06):100-109.

[54]王少然.生鲜农产品冷链物流需求预测研究[D].西安:西安工程大学,2017.

[55]王晓平,闫飞.基于GA-BP模型的北京城镇农产品冷链物流需求预测[J].数学的实践与认识，2019,49(21):17-27.

[56]张壹宁.江苏省水产品冷链物流需求预测研究[D].昆明:昆明理工大学,2017.

[57]朱毅丁,张云川,马云峰,等.基于CNN-LSTM-AM神经网络的多维长序列物流需求预测[J].物流科技,2024,47(18):49-56+64.

[58]王琰琰,任俊玲.基于GA-ACO-BP神经网络的日用消费品物流需求预测[J].北京信息科技大学学报(自然科学版),2024,39(01):91-98.

[59]徐青青,缪立新.区域物流发展及研究综述[J].物流技术,2006(04):1-3+11.

[60]刘婷.山东省物流需求组合预测方法及其应用研究[D].首都经济贸易大学,2016.

[61]李全喜,金凤花,孙磐石.区域物流能力与区域经济发展的典型相关分析—基于全国面板数据[J].软科学,2010,24(12):75-79.

[62]张云龙,刘茂.灰色GM(1,1)模型在火灾事故预测中的应用[J].南开大学学报(自然科学版)(01):13-17.

[63]邓华灿,陈松林.基于灰色序列GM(1,1)模型的建设用地预测[C]//2006’全国土地资源战略与区域协调发展学术研讨会.2006.

[64]秦璐,刘凯.基于产业结构的区域物流需求分析[J].物流技术,2006(07):4-6.

[65]肖红,夏如玉,王孝坤等.基于AOA-LSSVM模型的枢纽城市物流需求量预测[J/OL].重庆交通大学学报(自然科学版),1-7[2024-02-23].

[66]葛成丽.基于GM(1,1)-Stacking模型的长三角地区物流需求预测研究[D].安徽理工大学,2024.

[67]耿玥.江苏省区域物流需求预测模型及其应用研究[D].安徽理工大学,2022.

[68]Dong G, Wei W, Xia X, et al. Safety risk assessment of a Pb-Zn mine based on fuzzy-grey correlation analysis[J]. Electronics, 2020, 9(01): 130.

[69]Maharana K, Mondal S, Nemade B. A review: Data pre-processing and data augmentation techniques[J]. Global Transitions Proceedings, 2022.

[70]Tianqi Chen, Carlos Guestrin. XGBoost: A Scalable Tree Boosting System[C]. Knowledge Discovery and Data Mining, 2016.

[71]Prokhorenkova L, Gusev G, Vorobev A, et al. CatBoost: unbiased boosting with categorical features[J]. Advances in neural information processing systems, 2018, 31.

[72]马晓君,宋嫣琦,常百舒,袁铭忆,苏衡.基于CatBoost算法的P2P违约预测模型应用研究[J].统计与信息论坛,2020,35(07):9-17.

[73]Quinlan, J. R. (1984). Induction of decision trees. Machine Learning, 1(01),81-106.

[74]Quinlan, J. R. (1993). C4.5: Programs for machine learning. Morgan Kaufmann Publishers.

[75]Breiman, L. (2001). Random forests. Machine Learning, 45(01), 5-32.

[76]徐慧丽.Stacking 算法的研究及改进[D].广州:华南理工大学,2018.