为了利用信息系统对象在数据空间中分布,通过对对象的模糊聚类,计算每一类在坐标轴上的统计值。利用正态概率分布特性,引入重叠度和空隙度选取合适的概率,确定最优的区间端点,实现连续属性离散化。
In order to use the distribution of objects in information system, statistical values are computed on an axis by fuzzy clustering. By means of the normal distribution, overlap measure and gap measure are introduced, and then a probability is selected suitably, As a result, the terminals of data ranges are obtained optically to realize the discretization of continuous attributes successfully.