一种基于近邻规则的缺失数据填补方法

A Missing Data Imputation Method Based on Neighbor Rules

数据缺失是数据挖掘与分析过程中的常见问题,若直接删除含缺失的事例可能导致不可靠的决策。为此,针对缺失数据的填补问题,提出一种基于近邻规则的缺失数据填补方法。根据关联规则的后件数据项进行分类,计算分类后的规则项与缺失项集间的相似度,用最相似的规则项值填补缺失值。实验结果表明,该方法具有较高的填补正确率。

Data missing is a common problem in data mining and data analysis process, it can lead to reliable decision-making if it is deleted with the cases directly. An imputation method of solving the missing data is put forward, which is based on association rule. In this method, the rules are classified by the rules＇ consequent, and then calculate the similarity of constrained rules cases＇ items and missing cases＇ items, impute the missing value with the most similar rule＇s item. Experimental results show this method has higher imputation accuracy.