Abstract: Decision evolution sets is a theory to deal with evolution problem of decision rules in time series. Decision evolution sets transfer the focus from static decision information system to dynamic time series, which is a new decision research method to research the evolution regulations in decision information system following the time variation. At present, the standard structure of decision evolution sets has some questions, for example, less forecast attributes, bigger forecast angel. The three-way structure of decision evolution sets is a good way to improve the prediction accuracy, but the threshold α and β are fixed. However, in the time series data is constantly changing, fixed α and β cannot adapt to these changes. This paper uses game theory methods to adjust α and β to adapt to the changing in time series, and uses an example to show the adaption.

Key words: rough sets, granular decision, decision evolution, game of three-way structure

摘要： 决策演化集是处理决策规则在时间序列上的演化问题的理论。决策演化集将着眼点从静态的决策信息系统转移到动态的时间序列上，研究决策信息系统在随着时间变化时的演化规律，是一种新的决策研究方法。目前，在决策演化集的标准结构下存在着一些问题，例如预测得到的属性较少，预测夹角偏大等问题。决策演化集的三支结构在提高预测准确度方面是一个有效的方法，但其阈值α和β是固定的。然而，在时间序列下数据是不停变化的，固定的α和β并不能很好地适应这种变化。利用博弈论的方法来调整修改α和β使其适应决策信息系统在时间序列下的变化，并通过实例来演示这种调整。

关键词: 粗糙集, 粒度决策, 决策演化, 三支博弈