| Bayesian analysis represents uncertainty with probability and learning and inference are realized by probabilistic rules.The classification model based on this method is widely used in many areas with interpretable and high accuracy rate.With the rapid development of the China's economy,credit evaluation has become a topic of concern.According to the characteristics of credit evaluation data,this paper presents a Bayesian classification based on feature selection using Random Forest.The experimental results on the German dataset of UCI show that the feature selection using Random Forest not only simplifies the structure of the Bayesian classification model,but also makes the classification model has higher prediction accuracy.The main work and innovations of this paper are as follows:(1)In this paper,a Bayesian classification algorithm based on feature selection using Random Forest is proposed.Random Forest is a kind of intelligent learning algorithm which can tolerate noise and has high stability,and its algorithm can be used to select features based on the redundant or irrelevant attributes.Then think of the higher accuracy of Na?ve Bayes model,this paper constructs the Naive Bayesian classification model based on feature selection using Random Forest(RF-NB).(2)In consideration of the "independence hypothesis" of the Na?ve Bayes model is often not true,this paper adds some necessary boundary to each leaf node to express the dependencies among the various attributes.So this paper establishes the Tree Augmented Na?ve Bayes classification model based on feature selection using Random Forest(RF-TAN).(3)In order to verify the effect of RF-NB and RF-TAN model,the Bayesian classification model based on feature selection using Random Forest is applied to the German dataset to guide credit evaluation.The experimental results show that the classified effect of RF-NB and RF-TAN model are obviously better than NB and TAN model. |
PDF Full Text Request