Applying Rough Set Theory to Establish Artificial Neural Networks Model for Short Term Incidence Rate Forecasting

摘要

Choosing input variable and networks architecture are key processes for modeling short term incidence rate forecast by artificial neural networks, in this paper a method based on rough set theory is proposed to deal with them. In the proposed approach, the key factors that affect the incidence rate forecasting are firstly identified by rough set theory and then the input variables of forecast model can be determined. On the basis of the process mentioned above a set of influence rules can been obtained through reductive mining process of attributes and attribute values, then a neural networks of incidence rate forecast model is established on the rule set and BP-algorithm is adopt to optimize the networks. The method indicates that incidence rate forecast model can be established according some theoretical principles and avoiding blindness. A practical application is given at last to demonstrate the usefulness of the novel method.