An assessment framework and the efficiency validation on the discretization of the belief rule base system
Jianbin Sun, Leilei Chang, Yuejin Tan · 2013
Due to the inability of the belief rule base in handling continuous attributes with multi-values, the discretization of the continuous distribution function should be done. The information loss rate (ILR) and the information loss probability mass (ILPM) are calculated to represent the information loss in the discretization process. And a framework to assess continuous attributes based on RIMER is proposed. In the end, a numerical case is studied to validate the efficiency of the proposed framework and measurements.