A qualitative prediction model based on rough set for rainstorm applied to numerical prediction products

A new attribute reduction algorithm for decision table in rough set theory has been developed based on factor combined analysis method instead of using the common reduction method of Boolean function in traditional rough set theory, due to the fact that Boolean function reduction hasn´t considered the combination analysis of predictors. Using decreasing FAR calculation with Japanese model products, Chebyshev Polynomial Expression method and Empirical Orthogonal Function (EOF) method, predictand and predictors from model products in Europe center were computed, and a qualitative prediction model based on rough set and factor combined analysis method was established for rainstorm with heavy rainfall. Results show that the Threat Score (TS) of the modified model is higher than that of the traditional stepwise regression method in the qualitative prediction of rainstorm.