基于支持向量机的商业银行风险预警研究

发布时间：2018-04-24 17:45

本文选题：商业银行 + 风险预警　；参考：《天津财经大学》2015年硕士论文

【摘要】：本文首先从风险预警指标和预警方法的选择两个方面对国内外的有关商业银行风险预警研究的文献进行了系统的梳理,提出了本文的研究内容和方向——基于支持向量机的我国商业银行风险预警研究。然后介绍了我国商业银行面临的主要风险类型,并且从外部环境因素和商业银行自身内部经营情况两个方面分析了商业银行风险的成因,并且以此为基础建立了商业银行的风险预警指标体系,指标体系中包括了能够反映宏观经济环境和商业银行经营情况的指标。在构建商业银行的风险预警模型的过程中,首先采用因子分析法对所选择的样本银行进行分类,然后对他们进行分类预测。在利用支持向量机进行预测的过程中,考虑了多种参数优化方法,对比实证结果发现遗传算法和粒子群算法的分类预测准确率更高。本文的创新之处在于：(1)从导致商业银行产生风险的内在和外在两方面原因入手,综合商业银行所面临的资本充足程度、信用风险、盈利水平、流动性风险等主要问题,建立了一个完善的风险预警指标体系；(2)在对选择的银行进行分类时,采用的是因子分析的方法,计算出了银行总得分的风险阈值,并以此为基础对其进行分类,这样分类的结果更贴近于所选银行的整体情况；(3)本文在构建支持向量机模型时,对比了网格遍历法、遗传优化算法和粒子群优化算法,检验了遗传优化算法和粒子群优化算法在支持向量机的参数优化方面的优越性,为商业银行风险预警提供了一种有效的方法。
[Abstract]:From two aspects of the risk early-warning index and the selection of early warning methods, this paper systematically reviews the literature on the risk early warning of commercial banks at home and abroad, and puts forward the research content and direction of this paper - the risk early warning research of commercial banks in China based on support vector machine. Then it introduces the commercial banks in China. The main risk types are analyzed from two aspects of the external environmental factors and the internal management of commercial banks. On the basis of this, the risk early-warning index system of commercial banks is established. The index system includes the indicators that can reflect the macro economic environment and the business bank management. In the process of building the risk early warning model of commercial banks, we first classify the selected samples by factor analysis and classify them. In the process of forecasting with support vector machines, a variety of parameter optimization methods are taken into consideration, and the classification of genetic algorithm and particle swarm optimization is found in comparison with the empirical results. The innovation of this paper is more accurate. The innovations of this paper are: (1) starting with the internal and external reasons leading to the risk of commercial banks, the main problems such as capital adequacy, credit risk, profit level, liquidity risk and other major problems faced by commercial banks have established a perfect risk early warning index system; (2) When the selected banks are classified, the method of factor analysis is used to calculate the risk threshold of the total bank scores and classify them on this basis. The results of this classification are more close to the overall situation of the selected banks. (3) in this paper, the grid traversal method, the genetic optimization algorithm and the genetic optimization algorithm are compared when the support vector machine model is constructed. Particle swarm optimization (PSO) is used to test the superiority of genetic optimization algorithm and particle swarm optimization algorithm in support vector machine parameters optimization, which provides an effective method for risk early warning of commercial banks.

【学位授予单位】：天津财经大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F832.33

【相似文献】