中原工学院学报

会计核算数据存在连续变量离散化特性，在概念漂移情况下，难以捕捉数据中非线性连续变量交互效应，降低了数据分类的准确性。为此，提出一种基于决策树的会计核算数据结果自动分类方法。先通过获取会计核算数据集的统计概率密度特征量，捕捉数据的局部特征，实现连续变量的离散化处理。再对连续变量离散化结果进行模糊化处理，利用模糊信息增益最大值构建决策树，捕捉变量之间的非线性关系和交互效应，集成多棵决策树的分类结果，降低单一决策树的偏差，实现会计核算数据结果的自动分类。实验结果表明：所提方法能够高精度地处理会计核算数据，精准刻画其中的非线性关系。该方法满足实时性需求，为财务决策提供了及时、可靠的数据支持。

Accounting data has the characteristic of discretization of continuous variables.In the case of concept drift,it is difficult to capture the interaction effects of nonlinear continuous variables in the data,which reduces the accuracy of data classification. To address this,an automatic classification method for accounting data results based on decision trees is proposed. Obtain the statistical probability density feature quantities of the accounting data set,capture the local features of the data,and achieve the discretization processing of continuous variables. The discretization results of continuous variables are fuzzified. A decision tree is constructed by using the maximum gain of fuzzy information to capture the nonlinear relationships and interaction effects among variables. The classification results of multiple decision trees are integrated to reduce the deviation of a single decision tree and achieve the automatic classification of accounting data results.The experimental results show that the proposed method can process accounting data with high precision and accurately depict the nonlinear relationships therein. It meets the real-time requirements and can provide timely and reliable data support for financial decision-making.