Bound dynamic Bayesian network classifierfor enterprise operational risk evaluation

Abstract

The current risk evaluation method general presumes mutually independent indexes at the same level and realizes class evaluation by l weighing sum and interval division of the results. But in reality the indexes on the same level are not mutually independent and the relation bet different indexes is not linear. Through nodes ordering and local scoring, the thesis tries to set up the bound dynamic Bayesian network class Enterprise operational risk evaluation on the basis on such classifier can effectively avoid the aforesaid problems and provide new thoughts and met for intelligent and scientific risk evaluation of enterprise operation.