A GA-based approach to rough data model
Jinjie Huang, Shiyong Li · 2004
A genetic algorithm (GA) approach is presented to build the rough data model (RDM), which is a new methodology introduced by Kowalczyk in 1996 to deal with the inconsistence and uncertainty in database. Genetic algorithms (GAs) play two main roles in the proposed method: one is to select the best subset of the condition attributes, the other is to choose cut points from a candidate cuts set for discretization of the continuous valued attributes. The input space is then partitioned appropriately and a mapping relation between the input product subspaces and decision classes can be established. Moreover, a restricted genetic operator is designed for GAs to utilize the domain knowledge for faster convergence. Experimental results of two examples show the effectiveness of our approach.