Research On Early Warning And Control Strategy Of Overseas Foreign Direct Investment Risks Of Mineral Resource-based Enterprises Based On Multi-Classifier Fusion

With the continuous improvement of the competitiveness of Chinese enterprises,the slowdown of domestic economic growth,and the deepening of China’s reform and opening up,international cooperation,and especially the "Belt and Road" strategy,China’s outward foreign direct investment(OFDI)activities have become increasingly frequent.With the support and guidance of policies,more and more mineral resource-based enterprises(MREs)have joined the army of OFDI.However,due to the political,economic,and cultural differences between the host country and the home country,and the lack of experience of China’s MRES in the international mining and capital markets,and the fallacy of Western media’s prevailing "China’s rising threat theory",Outward foreign direct investment of mineral resource-based enterprises(OFDI-MREs)in China are facing a series of risks.The investment scale is large but the return on investment is very low.Many OFDI-MREs eventually fail.At present,the research on OFDI-MREs mainly focuses on the investee country,focusing on the impact of external risks such as politics,economy,law,and cross-cultural management on OFDI performance or conducting a qualitative analysis from the internal perspective of the enterprise.Financial analysis or machine learning methods warn of risks.However,the method of early warning of OFDI risks for different investment entities should be different,and their management and control strategies should be "medical to the situation".The purpose of this thesis is to realize OFDI-MREs risk warning and control.Based on the review of OFDI-related literature,the qualitative research model will be developed to the basis of theoretical knowledge and methodology such as management science,economics,computer science.Quantitative research model based on data mining and Multi-class fusion model(MCFM)promotes cross-disciplinary research methods and fusion of multi-source data and knowledge.Therefore,driven by the “Belt and Road” strategy,one of the focuses of future research will be investment in MREs,which has important theoretical significance and practical value for the faster and better development of China’s economy and society.Based on the risk mechanism of OFDI-MREs,this paper proposes a novel method—a comprehensive early warning of OFDI-MREs using a combination of Coefficient of variation method(C.V),system clustering and MCFM.First,we select logit regression(Logit),neural network(NN),decision tree(DT)and support vector machine(SVM)four single classifiers,and use Self-organizing Data Mining(SDM)technology to make decision-level fusion of the output of the single classifier to build a multi-classifier early warning model.Based on the risk generation mechanism of OFDI-MREs,an initial indicator system of risk early warning is constructed,and three Rate of Return on Common Stockholders’ Equity,Earnings Per Share and Capital Accumulation Rate are used as performance indicators;Secondly,we conduct empirical research.The experimental data comes from 173 samples of 42 MREs listed in China.First,the initial index system is simplified by the C.V method to obtain the final effective warning indicators.Then use systematic cluster analysis to divide OFDI risk into four levels.Finally,the indicator system and MCFM are used to realize the quantitative evaluation of OFDI-MREs risk.The model is continuously optimized based on experimental results and test results to obtain a fusion model and risk level early warning that meets the principle of optimal complexity.As a result,the current status of OFDI-MREs in China is explored.Finally,a case analysis is carried out,and based on the results of the OFDI-MREs risk early warning and the status quo in China,suggestions for risk early warning management and control strategies of OFDI-MREs are proposed.The main research results are as follows: First,a hierarchical early warning index system has been established.After reducing the index,it has 20 indicators in three dimensions.Second,according to the Return on Common Stockholders’ Equity,Earnings Per Share and Capital Accumulation Rate,the risks faced by OFDI-MREs are divided into four levels,and most OFDI-MREs are found to be at high risk.Third,the proposed Multi-class fusion model based on self-organizing data mining(MCFM-SDM)is better than four widely used single classifier models(logit,NN,DT,SVM)and four commonly used MCFM(Such as majority voting,Bayesian methods,and genetic algorithms)have higher accuracy and stability.Based on this,we discussed some suggestions for control strategies.Firstly,an OFDI case was used to analyze risk aversion.Secondly,the enterprise needed to establish a data collection system to lay a good foundation for the construction of enterprise big data applications.All relevant parties must start to establish a risk pre-control mechanism.Finally,based on the above research conclusions,from the perspective of MREs,we propose management and control strategies to reduce the risk of OFDI and enhance its risk prevention capabilities.In general,the research results of this thesis are of great significance in theory and practice.In theory,the OFDI-MREs risk assessment system was established,and the OFDI-MREs risk early-warning method was improved.In practice,finding out the pattern and cause of OFDI-MREs risk,providing guidance for policy formulation of China’s OFDI-MREs risk control strategy,and also having certain reference value for reducing the OFDI risk of other groups.The thesis include 16 figures,17 tables,114 references...