基于层次贝叶斯网络聚类的廉政风险评估分析

Evaluation Analysis of Integrity Risk Based on Hierarchical Bayesian Network Clustering

本文针对行政执法廉政风险的评估特点和现有评估方法存在的问题,建立了改进的层次贝叶斯网络聚类方法。这种方法对例子数量少、残缺甚至完全没有的特殊样本,也能够进行规则提炼和预测。基于改进的层次贝叶斯,并结合廉政风险的结构体系和历史数据,构建了廉政风险等级评估模型。经验证,预测准确率提升了32.03%,表明该模型预测结果准确、可靠,可用于行政执法廉政风险等级的评估。本研究对行政执法廉政风险评估理论具有一定参考和借鉴意义。

In view of the assessment characteristics of administrative law enforcement integrity risk and the problems existing in the existing assessment methods,an improved hierarchical Bayesian network clustering method was established.This method can also refine and predict the rules for special samples with a small number of examples,incomplete or even no at all.Based on the improved hierarchical bayesian and combined with the structural system and historical data,a integrity risk level evaluation model was constructed.And the prediction accuracy was showed to be increased by 32.03%with model verification.The prediction results of this model were accurate and reliable,and could be used to evaluate the integrity risk level of administrative law enforcement.This study had certain reference significance for the theory of integrity risk assessment of administrative law enforcement.