Sensitivity analysis of Bayesian networks to parameters of the conditional probability model using a Beta regression approach

作者：

Highlights：

• Global sensitivity analysis for BBN addressed with the viewpoint of beta regression.

• Applicable to any types of BBN (discrete, Gaussian, hybrid).

• Contributions are graphically depicted by the partial effects on the beta mean.

• Confidence measured by the evolution of the beta σ parameter.

• Combination with gradient boosting and stability analysis enhances the screening.

摘要

•Global sensitivity analysis for BBN addressed with the viewpoint of beta regression.•Applicable to any types of BBN (discrete, Gaussian, hybrid).•Contributions are graphically depicted by the partial effects on the beta mean.•Confidence measured by the evolution of the beta σ parameter.•Combination with gradient boosting and stability analysis enhances the screening.

论文关键词：Bayesian network,Sensitivity,Distributional regression,Beta distribution,Gradient boosting,Stability selection

论文评审过程：Received 22 March 2019, Revised 3 December 2019, Accepted 9 December 2019, Available online 10 December 2019, Version of Record 20 December 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2019.113130