The Patching Methods of the Bayesian Network Data
Gong Yishan, Hao Jia · 2009
In this paper, as the issue of Bayesian networks with missing data bring out a repair method of data. This method through a score function to score all nodes, so can obtained for all states points of the Bayesian network. This is to say all states which belong to Bayesian networks matching degrees can finalized determine the add data's states. Experiments show, this method can be learning Bayesian network structure inference or to repair data to make it a complete data it can effectively improve the utilization rate of missing data.