信息准确度在Bayes决策中的影响研究

得到 Bayes决策中抽样信息的最大期望收益值 EMV*s 的新计算公式 ,给出抽样信息的信息量定义 ,讨论了不同准确度抽样信息的条件熵、信息量和 EVSI的性质 ,得出完全信息的条件熵最小、信息量和 EVSI最大 ,均匀分布信息的条件熵最大、信息量和 EVSI最小的结论.

The Study of Influence of the Accuracy of Information in Bayesian Decision Analysis

In this paper, new computing formulas of maximal expected monetary value $EMV *-s$ are derived. The definition of quantity of sample information is given. The properties of conditional entropy, information quantity and EVSI of sample information with the distinct accuracy are discussed. Some conclusions are also obtained that the conditional entropy of complete information reaches the minimum,the quantity of sample information and EVSI the maximum, and that the condition entropy of uniform information reaches the maximum,the quantity of sample information and EVSI the minimum.