Study Of Variable Precision Rough Set Model And Its Application

The Variable Precision Rough Set Model (VPRSM) is a generalization of the Rough Set Theory (RST). It deals with partial classification by introducing a probability value β which denotes the proportion of correct classifications. Here, in this paper, the main concepts of VPRSM are introduced formally. Based on these, some important problems in VPRSM are focused.How to estimate the significance of attributes in decision table is a pop problem. Many literatures have already researched on it. while the researches are mostly based on the traditional RST. Here we bring forward a new method based on VPRSM and apply it to the combinational forecast.β and γ are two important parameters in VPRSM. β shows the proportion of correct classification and γ is defined as the quality of classification. We discuss the interrelation between the two symbols and give a β -domain concept transforming the traditional concept base on the value of β to the domain of β.The need for the reduct is important to the final objective of constructing a series of rules to classify the objects in the VPRSM. The traditional definition of β -reduct is keeping the same quality of classification. While we find sometimes the definition will cause abnormity. So according to the abnormity we show a new β -reduct definition. The core attributes is the base of the heuristic algorithms. In this paper, according to the problem of calculating the core attributes of an inconsistent decision table, two new algorithms are presented.Finally, we show a data mining model based on the VPRSM. And then we also analyse the WINE data sets of UCI database by the model. This experiment shows the model is useful.