【作者】 刘阳；

【导师】 肖毅；

【作者基本信息】 华中师范大学 ， 图书情报， 2017， 硕士

【摘要】 在深入推进“互联网+”行动和国家大数据发展战略的背景下,现代信息服务业迎来了高速发展的契机。上市公司是我国企业中具有代表性的一类,在其所处行业中有着重要的地位和影响力,随着行业发展不断深化,信息服务业上市公司在发展过程中也面临着一系列挑战。信息资源相对不足,投资分散且结构不合理,信息资源开发利用率低,服务范围狭窄等问题,现有产业的不断发展和客户对信息服务需求质量的不断提高,发展过程中存在的问题也日益凸显,风险水平日益提升。为了较好的适应不断变化的经营环境,运用适当的技术手段,构建财务风险预测模型是有实际意义的。本文一共分为六个部分。第一部分主要是确定了研究的背景和意义,总结学者们在信息服务、数据挖据和财务预测等领域所做的研究文献,明确文章的研究内容、思路和创新点;第二部分主要是对数据挖据和财务风险的基础理论部分进行介绍;第三部分,重点分析我国信息服务业上市公司的财务风险,阐释了信息服务行业所独特的特征和特殊的财务风险。第四部分主要为构建财务预测模型做数据准备和指标选取,选取了沪深两市信息服务业上市公司338家作为研究对象,从经过审计且公开的财务年报中获取数据,选取了 66家公司作为样本,从偿债能力、营运能力、盈利能力、成长能力、现金流量和公司规模6个大类,选出24个指标作为财务预测模型的指标体系。第五部分,为模型的构建和实证研究,应用R语言作为构建模型的工具,将清洗后的数据构建样本子集,应用SVM算法对样本做预测,最后给出预测的结果。第六部分为全文的总结和概况,提出研究的不足之处和展望。在模型的样本选择上,选取的是2012年到2016年之间,首次被标为ST的公司,共有17家,按照同大类同规模的原则,配对出17家非ST公司,再从正常样本中随机选出22家,共计66家样本。本文参考了国内外对财务预测的研究成果,梳理了大量的相关文献、总结了各种方法的优缺点之后选择用SVM进行建模。这是由于该算法对变量分布没有严格限制,兼具较好的学习能力和泛化能力,即使是在小样本情况下也有着较高的模型预测精度。实证结果表明,基于SVM构建的财务预测模型有较好的预测水平。更多还原

【Abstract】 Under the circumstance of "Internet+" and "Big Data",Modern Information Service Industries embrace a golden opportunity for rapid development. As a typical kind of enterprise in China, listed company has important status and influence in its field. With the deepen of its field,list of information service companies faces a series of challenges,lack of information resources, diversified and irrational investments,low exploitation and utilization ratio of information resources, small service area, etc. With the continuous development of industry and enhancement of service quality demand, the problems are highlighting and risks are increasing. The financial early forecasting model is beneficial to detect problems in management timely, thus strengthen the ability to resist risks and to accustom the varying management environment. Hence, it is necessary to forecast the financial risk of information service industry listed companies.This paper consists of 6 parts. The first part mainly gives the background of the research, concludes papers regarding information services, data mining and financial risk forecasting and clarifies the subject,methodology and innovations. The second part introduces knowledge about data mining and financial risk. The third part mainly analyze the financial risks of list of information service companies in China, defines the concept of information service industry and interprets the characteristics of this industry and its special financial risk. The fourth is mainly about model building and data selection. This part selects 338 listed companies from Shanghai and Shenzhen boards and obtain data from the annual financial reports. The sample consists of 66 companies. There are 24 indices in the financial early forecasting mode. The fifth is about empirical analysis. The last gives a conclusion of this paper, proposes the deficiency and expection.This paper chooses Support Vector Machine in model building after examining the regarding research findings. This is due to the unrestraint of variable and the generalization ability. This model can give precise results even under small sample. It should be noticed that, this paper selects 17 new ST companies during 2012 to 2016. In Accordance with the rule of the same sort and the same size, this paper pairs 17 non-ST companies. In order to keep the correspondence of the sample, 22 companies are chosen from the other companies. Hence,the sample consists of 66 companies in total. The predictability increases with the approach of the risk. This greatly enhances the practicability of this model and give support to managers in risk control.更多还原