欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
PD方法在收益管理无约束估计应用中较EM算法更加具有灵活性。无约束需求数据的非正态化假设更加符合收益管理实践中产品各＂预售提前期间隔＂的需求数据特征,包括对数正态、伽玛、指数和泊松分布。本文分别建立了上述非正态假设下的PD方法;并针对计算过程中可能出现的样本数据均受到截尾的情况,对所提方法的应用进行了改进;数值算例说明了所提方法迭代过程简单,且易收敛;通过与EM算法进行比较,验证了所提各PD方法的有效性,并就参数Υ对无约束估计效果的影响进行了分析和总结。
英文摘要：
The PD method is more flexible than EM algorithm for unconstraining estimation in revenue management. An assumption of Non-normal distribution for the nominal unconstrained demand data is more suitable for the demand characteristics in the Lead-time Intervals of the products under the practice of revenue management, including Lognormal, Gamma, Exponential and Poisson distribution. This paper derives formulae of PD methods when nominal demand is assumed to follow non-normal distributions mentioned above. Concerning the probability that sample data may be entirely censored in the calculation process, the applications of them are improved. Numerical examples are given to illustrate the simple iteration and the easy convergence processes of the proposed methods. Compared with the EM algorithms, the effectiveness of the proposed PD methods is verified. The impacts of parameter τ on the effect of unconstraining estimation are also analyzed and Smnlnarized.
同期刊论文项目
同项目期刊论文
期刊信息
位置：