基于内点法的稀疏逻辑回归财务预警模型

发布时间：2018-01-05 00:19

本文关键词：基于内点法的稀疏逻辑回归财务预警模型　出处：《计算机工程与设计》2013年06期 　论文类型：期刊论文

【摘要】：逻辑回归已广泛应用于财务危机建模,但是一定程度存在过拟合问题。为了避免建模出现上述问题,提出了基于L1正则化逻辑回归的财务预警模型。该模型是一种稀疏模型,能同时实现变量选择和参数估计,具有较强的鲁棒性。同时,针对L1正则化逻辑回归问题的求解,提出了一种高效的基于内点法的求解算法。结合沪深股市A股制造业上市公司进行实证分析,分析结果表明,L1正则化逻辑回归模型在预报精度、经济解释性等方面明显优于其他逻辑回归模型,并且提出的内点法与其它求解算法相比具有一定的优越性。
[Abstract]:Logistic regression modeling has been widely used in the financial crisis, but there is a certain degree of over fitting problem. In order to avoid modeling such problems, put forward the financial early-warning model of L1 regularization based on logistic regression. The model is a sparse model, can realize the variable selection and parameter estimation, and has strong robustness. At the same time, according to the to solve the L1 regularized logistic regression problem, an efficient algorithm was proposed based on interior point method. Combined with the Shanghai and Shenzhen stock market A shares of listed company of manufacturing industry in the empirical analysis, the analysis results show that the L1 regularized logistic regression model in prediction accuracy, the economic explanation is superior to other logistic regression model, and put forward the interior point method has certain superiority compared with other algorithms.

【作者单位】：华东交通大学信息工程学院;江西财经大学科研处;
【基金】：国家自然科学基金项目(61065003) 教育部人文社会科学研究规划基金项目(10YJC630379) 江西省自然科学基金项目(2010GZS0034)
【分类号】：F406.72;F832.51;TP18
【正文快照】： 0引言财务危机建模方法有很多,如单变量分析(univariateanalysis)[1]、线性判别分析(linear discriminate analysis,LDA)[2]、逻辑回归(logistic regression,LR)[3]、人工神经网络(artificial neural network,ANN)[4]和支持向量机(support vector machine,SVM)[5]等。基于逻

【参考文献】