中文摘要：

随着大数据时代的到来,企业的财务数据越来越复杂,在高维的财务数据中选出重要的影响因素,可以帮助企业人员对企业发生财务危机的可能性进行分析预测,降低企业的风险。研究企业财务危机的传统方法大多数是基于低维数据,而目前的财务数据变量众多。惩罚函数（penalty function）不仅能在高维数据的情况下进行稳定的变量选择,还能进行参数估计。采用基于Logistic回归惩罚函数模型研究企业财务危机预警问题,通过选取48个财务指标来进行建模和分析,找出影响企业财务危机的主要因素,并从预测精度、风险区分能力、变量选择这3个方面与传统方法进行对比,最终得到基于Logistic回归的惩罚函数方法更具有优越性。这对企业进行财务危机分析具有预警作用。

英文摘要：

As the company＇s financial data becomes more complex in. the era of large data, the im- portant influencing factors choosed from high-di-mensional financial data is increasingly playing an. significant role in. the financial analysis, which can. help enterprises to analyze and forecast the possibility of financial crisis. But at present, most of the traditional methods of studying corporate financial crises are based on. low-dimensional da-ta. The penalty function, can. not only stabilize the variable selection, in. high dimensional data, but also estimate the parameters. In. this paper, lo-sing the logistic regression penalty fimction model to study the corporate financial crisis early warningproblem, selecting 48 financial indicators to modeling and analysis and finding out the main, factors influencing the financial crisis of the enterprise . The penalty function, method is compared with the traditional method from the prediction, accuracy, risk differentiation, ability and variable selec-tion.. Finally, Logistic regression, based on. the penalty function, of the financial crisis analysis of enterprises with early warning role is more superior.