基于内点法的稀疏逻辑回归财务预警模型

Sparse logistic regression based on interior-point method for financial early-warning model

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。

Logistic regression （LR） is widely applied in building financial early-warning model, but to a certain extent, it has o ver-fitting problem. To avoid this problem, the financial early-warning model based on L1 regularized logistic regression is put forward. This model is a sparse model, can select variables and estimate coefficients simultaneously, and has strong robustness. At the same time, aiming at the solving of L1 regularized logistic regression problem, an efficient algorithm based on interiorpoint method is presented. Experiments are implemented on the financial data of A share manufacturing listed companies of the Shanghai and Shenzhen stock markets. The experimental results show that L1 regularized logistic regression model is apparently superior to other LR model in the accuracy of prediction, the economic interpretation, etc. Moreover, compared to other algo- rithms, the proposed interior-point algorithm has some superiority.