论文部分内容阅读
为了降低单一分类模型的不稳定性,提高个人信用评估的准确性,提出一种基于多分类器的组合分类的个人信用评估模型。首先进行特征划分,将数据样本的特征集合分为多个特征子集,然后再在各个特征子集上分别建立不同的分类模型。在分类模型结果组合阶段,对各个特征子集的分类模型的结果进行组合从而形成最终分类结果。实证研究以决策树,朴素贝叶斯以及支持向量机作为基分类器,通过德国银行数据集上的比较,说明提出的方法相对于单一分类方法具有更高的分类准确性,可以应用于个人信用评估。
In order to reduce the instability of single classification model and improve the accuracy of personal credit evaluation, a personal credit evaluation model based on multi-classifier combination classification is proposed. Firstly, feature classification is used to divide the feature set of data samples into multiple feature subsets, and then different classification models are established on each feature subsets. In the stage of the combination of classification model results, the results of the classification model of each feature subset are combined to form the final classification result. Empirical studies using decision trees, naive Bayes and support vector machines as the base classifier, through the comparison of data sets on the German bank, shows that the proposed method has higher classification accuracy than the single classification method and can be applied to personal credit Evaluation.