关键词: 混合模型, 粗糙集, 多准则库存分类, 决策规则

Abstract: A predictive approach of hybrid model for multi-criteria classification of inventory items is presented in this paper. The whole procedure includes two parts as follows. The first is to apply the advantages of rough sets (RS) to handle the qualitative criteria and realize the reduction of multi-criterion attributes without information loss during inventory classification. Then, based on the defined three information probability rules, the exhaustive algorithm and the covering algorithm are developed to extract the classification decision rules. The second is to further classify the items in the inventory classification, which cannot be classified correctly by rough sets by artificial neural network (ANN) with BP algorithm. The proposed method can not only overcome the shortcomings of input limitation in ANN, but further improve the prediction accuracy.Finally, three comparable simulation experiments for the real inventory classification problem have been performed using the ANN model with the BP algorithm and the rough sets method. It is proved that the use of the hybrid model is a promising analytical tool in the real decision-making systems.

Key words: hybrid model, rough sets, multi-criteria inventory classification, decision rules