CAJ | 학술논문

国家社科期刊

系统工程理论与实践 계통공정이론여실천
Systems Engineering—Theory & Practice
2008年 6期 129~136 ,共null页

李旭升郭春香郭耀煌李旭升郭春香郭耀煌

리욱승 곽춘향 곽요황

信用评估贝叶斯网络树增强朴素贝叶斯分类模型扩展的树增强朴素贝叶斯分类模型神经网络分类模型信用評估貝葉斯網絡樹增彊樸素貝葉斯分類模型擴展的樹增彊樸素貝葉斯分類模型神經網絡分類模型
신용평고 패협사망락 수증강박소패협사분류모형 확전적수증강박소패협사분류모형 신경망락분류모형
credit scoring; bayesian network; tree augmented naive Bayesian classifier; extended tree augmented naive Bayesian classifier; neural network classifier

针对信用评估问题的特点，在推导混合数据极大似然函数的基础上，提出了扩展的树增强朴素贝叶斯网络信用评估模型，用10层交叉验证在真实数据集上进行了测试并与神经网络分类模型进行了比较．测试结果表明扩展的树增强朴素贝叶斯网络信用评估模型具有较高的分类精度，在信用评估中具有优势．
침대신용평고문제적특점，재추도혼합수거겁대사연함수적기출상，제출료확전적수증강박소패협사망락신용평고모형，용10층교차험증재진실수거집상진행료측시병여신경망락분류모형진행료비교．측시결과표명확전적수증강박소패협사망락신용평고모형구유교고적분류정도，재신용평고중구유우세．
Aiming to the specialty of credit scoring, the maximum likelihood function of hybrid data is deduced, and a new credit scoring model called Extended Tree Augmented Naive Bayesian Classifier is put forward. The proposed model is tested using 10-fold cross validation with a real world data set, and compared with neural network models. Results demonstrate that the credit scoring model on extended tree augmented naive Bayesian classifier is competitive with neural network models and predominant in credit scoring domain