In this paper, we report our success in building efficient scalable classifiers in the form of decision tables by exploring capabilities of modern relational database management systems. In addition to high classification accuracy, the unique features of the approach nclude its high training speed, linear scalability, and simplicity in implementation. More mportantly, the major computation required in the approach can be implemented using standard functions provided by the modern relational DBMS. This not only makes implementation of the classifier extremely easy, further performance improvement is also expected when better processing strategies for those computations are developed and implemented in RDBMS. The novel classification approach based on grouping and counting and its implementation on top of RDBMS is described. The results of experiments conducted for performance evaluation and analysis are presented.
展开▼
