Improved algorithm for relational decision tree classification

Many relational classification methods, which are practically applicable for discovering knowledge in relational databases, are proposed recently. Decision tree induction is one of the major approaches to classification, so upgrading this approach to a relational setting has thus been of great importance. RDC is one of the algorithms that upgrade decision tree to relational database. It uses Class Label propagation to obtain the information that used in the construction of decision tree. However, it can achieve more efficient and accuracy classification result, RDC has its own drawbacks. In this paper we propose an improvement of the RDC algorithm, which using a more complex method for dealing with missing values to obtain higher classification accuracy.