一种基于约束的空缺数据预处理算法

A Preprocessing Algorithm for Missing Data Based on Constraints

现实数据库中的数据经常包含空缺数据,这些空缺数据的存在可能导致数据挖掘和基于数据仓库决策的失败。该文针对一类包含聚集信息的空缺数据,给出了一种基于约束的预处理算法。文章首先描述了算法的处理思想和步骤,然后给出了实验结果。分析和实验表明,算法是可行的并且能够比不使用约束信息的方法更精确地重构空缺值。

Real-world databases often contain missing d ata,which can make data mining and decision based on data warehouse failure.Th is paper presents a constraints-based preprocessing algorithm for a kind of mis sing data which contain aggragation information.Firstly the idea and the proces s steps are described,then the experimental result is given.Analysis and exper iment indicates that the algorithm is feasible and can reconstruct the misising data more accuracy than the method which does not use constraints information d oes.