Classification Based on Attribute Dependency

The decision tree learning algorithms, e.g., C5, are good at dataset classification. But those algorithms usually work with only one attribute at a time. The dependencies among attributes are not considered in those algorithms. Thus, it is very important to construct a model to discover the dependencies among attributes and to improve the accuracy of the decision tree learning algorithms. Association mining is a good choice for us to concern with the problems of attribute dependencies. Generally, these dependencies are classified into three types: categorical-type, numerical-type, and categorical- numerical-mixed dependencies. This paper proposes a CAM (Classification based on Association Mining) model to deal with such kind of dependency. The CAM model combines the association mining technologies and the traditional decision-tree learning capabilities to handle the complicated and real cases. According to the experiments on fifteen datasets from the UCI database repository, the CAM model can significantly improve both the accuracy and the rule size of C5. At the same time, the CAM model also outperforms the existing association-based classification models, i.e., ADT (Association-based Decision Tree) and CBA (Classification Based on Association).