Modeling and evaluation of the efficiency of policy for regional revitalization by network data envelopment analysis

在本研究中，我们介绍了两种类型的网络DEA模型，即加法模型和极差调整度量(RAM)模型，并研究了这两种模型的一些性质；网络DEA模型可以处理中间产品。加法模型和RAM网络DEA模型都是非径向模型，适用于输入或输出发生非比例变化的问题，并且具有平移不变性的性质，利用这一性质，我们可以处理某些输入或输出中的负数据。这些模型不仅可以应用于具有内部或关联活动的具体问题(例如区域振兴问题)，而且可以将损失和利润视为输出。此外，我们得到了强代表给定递归离散决策过程的序列决策过程子类存在的充要条件，并给出了它们的几个算法，如果我们将它们的使用限制为求解具有附加代价的最短或最长路径问题，则可以称为广义Dijkstra方法，因为它们与Dijkstra方法一致。从这些结果中，我们可以明确网络DEA在区域振兴问题建模中需要考虑的问题。

In this research, we have introduced two types of network DEA models, Additive model and Range Adjusted Measure (RAM) model, and investigated several properties of the models ; Network DEA models can deal with intermediate products. The Additive model and RAM Network DEA model are non-radial models which are applicable to some problems in which inputs or outputs undergo non-proportional changes, and have a property known as "translation invariance" ; from this property, we can deal with negative data in some of the inputs or outputs. These models can be applied to several concrete problems (for example, regional revitalization problem) not only with internal or linking activities but also with losses as well as profits treated as output.Moreover, we have derived necessary and sufficient conditions for the existence of the subclasses of sequential decision processes, strongly representing a given recursive discrete decision process, and presented several algorithms for them, which may be called generalized Dijkstra's method because they are coincident with Dijkstra's method if we restrict the use of them to solving shortest or longest path problem with additive cost measure. From these results, we can clear up issues to be taken into account in Modeling of the regional revitalization problem by network DEA.