Most of the KDD algoithms proposed in the literature have been applie to relatively small datasets and do not permit any integration with a DBMS. Hence, the application of these algorithms to the huge amounts of data found in current databsases and data warehouses faces seriious scalability problems, particularly the porblem of excessive learning tiem. THis paper investigates a way of improving the scalability of KDD algorithms, via discretization of ordinal or continuous attributes. This work has two novel aspects. First, we map a genric discretization primitive into an SQL query. Second, we propose a new discretization algorithm for classification tasks. We show how the new discretization algorithm can be implemented with good effect via the SQL primitive.
展开▼
