【摘要】 为了提高信息系统的相容性、解决离散化过程中阈值以难确定的问题,提出了一种改进的连续属性离散化算法.该算法以信息增益作为属性重要性的度量方法,以基于最小描述长度原理的决策系统信息熵作为离散化过程的评价函数,能够有效地提高离散化速度和精度,并且增强了系统的鲁棒性.仿真结果证明,此算法能够得到满意的离散化结果,是一种能够自动调节阈值的有效算法.更多还原

【Abstract】 In order to increase compatibility of information systems,given a new continuous attribute discretization arithmetic.The arithmetic use information gain as measure of attributes import and information entropy as estimate function of discretization course.It can enhance effectively discretization speed and precision,and gather head systemic robustness.The simulation result shown that the model has good coincidence with user’s subjective request and objective reality.更多还原