制造业上市公司营运资金风险预警对比研究

发布时间：2018-07-14 13:57

【摘要】：随着我国市场经济建设的完善,我国各行业迅速发展,各项经济指标每年以指数倍数增长,尤其是制造业已成为支撑我国GDP的主要产业之一,我国也成为制造业大国,在国际上也有“世界工厂”的称号。但是,由于能源、环境、人口等各项限制条件,中国的经济增长趋势放缓,究其原因,为了响应创新型国家建设的步伐,我国制造业逐步由“世界工厂”向高级技术创新类产业过度,转型的过程中企业生产经营活动都会面临困境,尤其是和日常运作相关的营运资金的管理和控制,有必要投入更多的精力关注因营运资金运作不当而带来的财务风险。因此为了制造业的稳步改革发展,需要准确的对制造业上市公司的营运资金风险进行预警研究。本研究在前人研究营运资金风险的基础上,首先,阐述营运资金风险预警和数据挖掘技术的相关理论基础,分析数据挖掘方法在预警分析中的优势和可行性；其次,选取两市A股被预警处理的制造业为研究对象,并在指标体系的选择准则的基础上,构建营运资金风险预警指标体系,并详细描述数据挖掘中BP神经网络、Logistic回归分析、C5.0决策树三种预测方法的基本原理和模型的构建；最后,结合SPSS Clementine运行程序,对选取的36家制造业样本公司进行营运资金风险预警实证对比分析,并对三种预测方法进行模型精确度评价,得到对比分析后的研究结论。通过研究对比分析得知,基于数据挖掘的上市公司营运资金风险预警模型具有很强预警能力；并且三种预测模型越靠近被预警处理的年限,预测精度越高,表明了上市公司营运资金风险是一个动态的变量,预警模型也有很强的时效性；通过纵向比较分析归纳出,本研究建立的预测模型中BP神经网络模型最好,Logistic回归模型最差,C5.0决策树模型居中；数据挖掘方法中,以知识发现为理论基础的模型预测精度较高,优于以统计分析为基础的预测模型。因此数据挖掘技术在营运资金风险预警分析中具有可行性,企业完全可以应用数据挖掘技术,进行营运资金风险管理决策,以提高资金使用效率。
[Abstract]:With the perfection of market economy construction in our country and the rapid development of various industries in our country, every economic index is increasing exponentially every year, especially the manufacturing industry has become one of the main industries supporting our GDP, and our country has also become a big manufacturing country. There is also the title of "World Factory" in the world. However, due to the restrictions on energy, environment and population, China's economic growth trend has slowed down. In order to respond to the pace of building an innovative country, The manufacturing industry of our country has gradually changed from "world factory" to "advanced technology innovation industry". During the process of transformation, enterprises will face difficulties in their production and operation activities, especially the management and control of working capital related to daily operation. It is necessary to devote more attention to the financial risks caused by the improper operation of working capital. Therefore, for the steady reform and development of manufacturing industry, it is necessary to study the working capital risk of listed manufacturing companies accurately. On the basis of the previous research on working capital risk, firstly, the paper expounds the relevant theoretical basis of working capital risk early warning and data mining technology, and analyzes the advantages and feasibility of data mining method in early warning analysis. Based on the selection criteria of the index system, the working capital risk early warning index system is constructed. The basic principle and model construction of three prediction methods of BP neural network and logistic regression analysis and C5.0 decision tree in data mining are described in detail. Finally, combined with SPSS Clementine running program, The empirical comparative analysis of working capital risk early warning is carried out on 36 manufacturing industry sample companies, and the model accuracy of three forecasting methods is evaluated, and the conclusion is obtained. Through comparative analysis, we know that the working capital risk early-warning model of listed companies based on data mining has strong early-warning ability, and the closer the three forecasting models are to the years of early warning, the higher the prediction accuracy is. It shows that the working capital risk of listed companies is a dynamic variable, and the early warning model has strong timeliness. In the prediction model established in this study, the BP neural network model is the best and the logistic regression model is the worst C5.0 decision tree model, and in the data mining method, the prediction accuracy of the model based on knowledge discovery is higher than that of the model based on knowledge discovery. It is superior to the prediction model based on statistical analysis. Therefore, data mining technology is feasible in the early warning analysis of working capital risk. Enterprises can use data mining technology to make working capital risk management decision, in order to improve the efficiency of capital use.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F406.7;F425

【参考文献】