A recycle technique of association rule for missing value completion
Jau-Ji Shen, Ming‐Tsung Chen · 2003
The technologies of internetworking and databases have expedited the distributed database systems being implemented in the global information systems. But we cannot avoid the occurrence of random missing values in the process of gathering data. The research on association rules can effectively establish the relationship of attributes in databases containing missing values. While this technique is used to resolve the issues on replacing the missing values, the integrity of the methodology and the measurability of the quality can already be verified in related essays. However, the accuracy of replaced missing values in the database can only be satisfactory with specific criteria and a higher completion rate cannot yet be achieved. In other words, a given accuracy cannot be reached at the same time as resolving the database validity issues. Therefore, we propose a rule recycle technique, which reuses and composes the rules in order to retrieve more complete attribute value association rules. This will not only enable the database recovered to improve the accuracy and completion rate but also increase the validity of missing value completion.