一种基于Boosting的集成学习算法在银行个人信用评级中的应用

Application on Individual Credit Score for Bank of a Boosting-based Ensemble Learning Algorithm

本文针对银行个人信用数据的分类预测问题,从数据集的特征选择和集成学习两个角度出发,提出了PCA-AdaboostLogistic集成学习算法。在采用Accuracy和AUC作为分类模型评价指标的前提下,本文选取了源于澳大利亚某银行的个人信贷数据集进行测试。测试结果表明本算法在有效提取关键特征后提高了Adaboost的稳定性,并且在分类准确度上相比单纯使用Logistic分类器有不同程度的提高。

This paper focused on classification prediction problem of the bank personal credit data,proposed a PCA-Adaboost-Logisticensemble learning algorithm based on feature selection and ensemble learning.Accuracy and AUC were used as the classification modelevaluation metric under the premise,this paper used the credit data sets from Australian banks to test the proposed algorithm.The resultsshow that the proposed algorithm improves the stability of the Adaboost after extract the key features,and the classification accuracy ishigher than the Logistic classifier.