基于机器学习的“一带一路”投资国别风险预测研究  

Investment Country Risk Prediction of“the Belt and Road”Based on Machine Learning

作　　者：向鹏成[1,2,3] 高天 段旭[1] 李东[1] Xiang Pengcheng;Gao Tian;Duan Xu;Li Dong(School of Management Science&Real Estate,Chongqing University,Chongqing 400045,China;International Research Center for Sustainable Built Environment,Chongqing University,Chongqing 400045,China;Construction Economics and Management Research Center,Chongqing University,Chongqing 400045,China)

机构地区：[1]重庆大学管理科学与房地产学院,重庆400045 [2]重庆大学可持续建设国际研究中心,重庆400045 [3]重庆大学建设经济与管理研究中心,重庆400045

出　　处：《工业技术经济》2024年第7期150-160,共11页Journal of Industrial Technological Economics

基　　金：中央高校基本科研业务费资助(项目编号:2022CDJSKPT25)。

摘　　要：“一带一路”倡议提出十年间,中国对沿线国家的投资规模持续扩大。然而,企业在抓住机遇,进行“一带一路”沿线国家投资的同时,也需要重点关注“一带一路”投资国别风险。本文从政治、经济、社会和对华关系4个维度构建“一带一路”投资国别风险预测指标体系;运用灰色关联分析计算样本国家的综合风险评价值;基于2012~2022年间“一带一路”沿线国家的数据,利用机器学习构建GA-BP神经网络、支持向量回归和随机森林3种预测模型;通过对比预测精度,确定最佳预测模型,利用2021年的指标数据,对2022年的投资国别风险进行预测。研究结果表明:(1)在“一带一路”投资国别风险的研究背景下,支持向量回归模型预测效果最优,证明机器学习模型能够有效应用于风险管理领域;(2)“一带一路”投资国别风险存在明显的地区差异,中东欧地区和东南亚地区投资国别风险普遍较低,而南亚地区投资国别风险普遍较高,但都存在特例。本文研究结果可为“走出去”企业在“一带一路”沿线国家的投资决策提供参考。China's investment in the nations along the Belt and Road initiative has increased significantly in the ten years since its inception.However,while enterprises seize the opportunity to invest in countries along“the Belt and Road”,they also need to focus on the country risks of“the Belt and Road”investments.This study constructs a country risk prediction index system for“the Belt and Road”investment from four dimensions:political,economic,social and relationship with China;calculates the comprehensive risk evaluation value of the sample countries by using grey correlation analysis;based on the data of the countries a⁃long“the Belt and Road”during the period of 2012~2022,the risk evaluation value of the countries along“the Belt and Road”is calculated by using grey correlation analysis.Based on the data of the countries along“the Belt and Road”between 2012 and 2022,three kinds of prediction models are constructed using machine learning:GA-BP neural network,support vector regression and random forest;the best prediction model is determined by comparing the prediction accuracy;and the country risks in 2022 are predicted using the indicator data of the 2021.The study's findings demonstrate that:(1)the machine learning model may be used successfully in the field of risk management since the support vector regression model performs best when considering“the Belt and Road”investment country risk.(2)The country risks associated with“the Belt and Road”investments vary significantly depen⁃ding on the region.While there are occasional exceptions,the country risks associated with investments in South Asia are generally higher and those in Central and Eastern Europe and Southeast Asia are generally lower.The study's findings can serve as a guide for“going out”businesses in“the Belt and Road”countries as they make investment decisions.

关 键 词：“一带一路”投资 国别风险 机器学习 风险预测 GA-BP神经网络 支持向量回归 随机森林 地区差异 

分 类 号：F752.7[经济管理—国际贸易] F224[经济管理—产业经济]