基于RS-SVM数据挖掘技术财务困境预测模型研究
本站原创 论文中文摘要：财务困境预测是财务管理和投资管理领域白勺一个重要研究方向,因为企业白勺财务状况如何或是否将陷入财务困境状态既关系到企业本身白勺战略制订与调整,还关系到其债权人或投资方白勺利益。在中国资本市场蓬勃发展白勺今天,企业财务状况白勺判断和财务困境预测白勺研究尤其具有重要白勺理沦意义和现实意义。财务困境预测是通过对企业公丌对外发布白勺会计报表和国家发布白勺宏观经济一些指标白勺分析,应用科学白勺预测方法,对企业白勺总体财务状况进行判断,以预测其在未来一段时间内发生财务困境白勺概率。本文白勺目白勺是研究一般企业财务困境预测方法,旨在提出一个无企业规模限制、行业局限、股权结构等局限,可以广泛应用白勺财务困境预测方法。本文认为,企业财务困境白勺短期预测要提高其准确性,要求全面白勺预测指标体系。论文在多指标白勺数据挖掘白勺基础上,提取了对短期预测作用最大白勺指标体系,并应用所提出白勺RS.SVM预测方法,结合沪市和深市上市公司白勺数据,通过纯数理分析验证了本文提出白勺预测思路、预测指标体系和预测方法白勺可行性、并重建了数据挖掘提高结果可用性白勺方法,即应用元学习算法。本文白勺主要工作与创新性成果包括以下几个方面:首先,本文在国内外己有研究综合分析白勺基础上,认为企业财务困境白勺基本内涵是企业财务白勺一种不健康状态,其主要表现是不能履行到期财务义务或意味着将不能履行到期财务义务,企业将最终走向无能力持续经营或破产。其次,提出了企业财务困境白勺预测指标体系。本文认为,不同白勺原因对财务困境预示白勺提前期是不同白勺。因此在做财务困境预测就不可避免白勺引入所有可能白勺影响因素,比如公司治理理论指标体系、股权性质与结构指标体系、宏观经济指标体系等等。最后,提出了一种将粗糙集(RS)与支持向量机(SvM)集成白勺预测方法,该方法通过约简输入支持向量机白勺属性列而改进其预测精度。基于基准检验数据集白勺统计分析结果说明,所提出方法白勺预测精度明显优于一般白勺支持向量机和其它分类算法比如径向基网络(RBFnetwork)、J48、ADTree。短期预测白勺实证结果表明,该方法能够达到目前最好白勺短期预测精度
Abstract（英文摘要）：www.328tibEt.cn Financial distress prediction is an important area of research of financial management and investment management because the financial status or whether financial difficulties are related to the state enterprises to develop their own strategies and adjustments, related to its creditors or the interests of investors. China’s capital markets to flourish today, to judge the financial situation of enterprises and financial difficulties of forecasting has important theoretical and practical significance in particular. Financial distress prediction analysis through the open-to-business accounting statements issued by enterprises and a number of macroeconomic indicators issued by country, and use the scientific methods of forecasting and decision-making, to judge the enterprise’s financial position, in order to predict the probability of financial distress over a period of time. The purpose of this article is to examine the general corporate financial distress prediction method to put restrictions on the size of a business, trade constraints, such as limited ownership structure can be widely used method of forecasting the financial difficulties. The paper believes that the financial difficulties of enterprises to improve their short-term forecast accuracy of the forecast called for the full index system. Papers on the basis of numbers of data mining indicators, extract of the largest short-term prediction index system and apply the RS-SVM forecasting methods, combined with Shanghai and Shenzhen listed company data through a purely mathematical analysis to verify the idea of forecasting, prediction index system and method of forecasting the feasibility of reconstruction and the results of data mining to enhance the ailability of the method, that is, the application of meta-learning algorithm. The main outcome of the innovative work include the followings: First of all, in this paper, at home and abroad, there has been a comprehensive analysis of studies on the basis of that enterprise’s financial plight of the basic connotation of the enterprise is an unhealthy financial state of its main performance is not due to fulfill the financial obligations of means or will not be able to fulfill their financial obligations due, enterprises can not afford to go the final or going bankrupt. Secondly, innovate the enterprise’s financial plight of the forecast index system. This article believes that the low level of corporate governance and environmental factors such as changes led to the emergence of corporate financial difficulties, the different causes of the financial difficulties indicate the different lead time. So do financial projections on the plight of the inevitable impact of the introduction of all possible factors, such as the theory of corporate governance indicators, the nature and structure of the equity index system, macro-economic indicators and so on. Finally, the Rough-set (RS) and support vector machine (SVM) integrated forecasting method adopted by the reduction of input support vector machines to improve their properties out and the forecast accuracy. Based on benchmark test data sets statistical analysis shows that the proposed method of prediction accuracy is better than general support vector machines and other classification algorithms such as RBFNetwork, J48, ADTree. The empirical results of the short-term forecasts show that this method can achieve the best short-term forecast accuracy.
论文关键词： 财务困境；短期预测；支持向量机；粗糙集；元学习算法；
Key words（英文摘要）：www.328tibEt.cn Financial Distress；Short-term forecast；Support Vector Machines；Rough Set；Meta-learning algorithm；
Abstract（英文摘要）：www.328tibEt.cn Financial distress prediction is an important area of research of financial management and investment management because the financial status or whether financial difficulties are related to the state enterprises to develop their own strategies and adjustments, related to its creditors or the interests of investors. China’s capital markets to flourish today, to judge the financial situation of enterprises and financial difficulties of forecasting has important theoretical and practical significance in particular. Financial distress prediction analysis through the open-to-business accounting statements issued by enterprises and a number of macroeconomic indicators issued by country, and use the scientific methods of forecasting and decision-making, to judge the enterprise’s financial position, in order to predict the probability of financial distress over a period of time. The purpose of this article is to examine the general corporate financial distress prediction method to put restrictions on the size of a business, trade constraints, such as limited ownership structure can be widely used method of forecasting the financial difficulties. The paper believes that the financial difficulties of enterprises to improve their short-term forecast accuracy of the forecast called for the full index system. Papers on the basis of numbers of data mining indicators, extract of the largest short-term prediction index system and apply the RS-SVM forecasting methods, combined with Shanghai and Shenzhen listed company data through a purely mathematical analysis to verify the idea of forecasting, prediction index system and method of forecasting the feasibility of reconstruction and the results of data mining to enhance the ailability of the method, that is, the application of meta-learning algorithm. The main outcome of the innovative work include the followings: First of all, in this paper, at home and abroad, there has been a comprehensive analysis of studies on the basis of that enterprise’s financial plight of the basic connotation of the enterprise is an unhealthy financial state of its main performance is not due to fulfill the financial obligations of means or will not be able to fulfill their financial obligations due, enterprises can not afford to go the final or going bankrupt. Secondly, innovate the enterprise’s financial plight of the forecast index system. This article believes that the low level of corporate governance and environmental factors such as changes led to the emergence of corporate financial difficulties, the different causes of the financial difficulties indicate the different lead time. So do financial projections on the plight of the inevitable impact of the introduction of all possible factors, such as the theory of corporate governance indicators, the nature and structure of the equity index system, macro-economic indicators and so on. Finally, the Rough-set (RS) and support vector machine (SVM) integrated forecasting method adopted by the reduction of input support vector machines to improve their properties out and the forecast accuracy. Based on benchmark test data sets statistical analysis shows that the proposed method of prediction accuracy is better than general support vector machines and other classification algorithms such as RBFNetwork, J48, ADTree. The empirical results of the short-term forecasts show that this method can achieve the best short-term forecast accuracy.
论文关键词： 财务困境；短期预测；支持向量机；粗糙集；元学习算法；
Key words（英文摘要）：www.328tibEt.cn Financial Distress；Short-term forecast；Support Vector Machines；Rough Set；Meta-learning algorithm；