基于属性依赖度的缺失值顺序填充算法

A SEQUENTIAL FILLING ALGORITHM FOR MISSING VALUES BASED ON ATTRIBUTE DEPENDENCY

提出一种基于属性依赖度的对同一元组中多个确失值进行顺序填充的方法。该方法首先利用粗糙集理论对属性进行约简,利用改进的ID3算法建立带有空分支结点的填充树,然后利用填充树对同一元组中多个缺失值按依赖度从大到小的顺序进行填充。实验结果表明,该算法在填充效果上明显优于现有的其他填充算法。

We bring forward a method which fills in order the missing values of the same tuple based on attribute dependency. First, the method reduces the attributes with rough set theory, and establishes the filling tree with empty branch nodes by the improved ID3 algorithm. Then it uses the filling tree to fill the missing values of the same tuple in descending order according to the dependency. Experimental results show that the proposed method outperforms the existing algorithms in filling effect.