残缺数据下动态随机算法及应用

Dynamic Stochastic Algorithm Based on Incomplete Data and Its Application

针对指标数据残缺的动态评价问题,提出了一种基于残缺数据的动态随机算法。首先依据时间维度上的分布情况将残缺值分为两类:离散型与连续型,并提出了对应的补足方法;然后在数据补足完整的基础上,利用随机模拟技术,计算优胜度矩阵,并推导出被评价对象之间的可能性排序。该算法避免了评价对象之间排序的绝对性,在对实际问题的解释方面具有较大弹性。最后,通过一个算例对该算法进行详细说明。

To the dynamic evaluation problem with incomplete data,this paper proposes a dynamic stochastic algorithm based on incomplete data.Firstly,the missing data are classified into two types,the discrete missing data and the continuous missing data.According to their distribution in time dimension,the relative filling methods are proposed in this paper.Then,based on complete data,the winning probability matrix representing the relative performance among alternatives is calculated,combining with the stochastic simulation,from which the probability sort between the alternatives is induced.The algorithm avoids the absolute sort results among alternatives and has more advantages in explanations for real-world applications.Lastly,the algorithm is illustrated in an example in detail.