To make rough set theory can deal with incomplete information system with noise effectively, combining the
generalized rough set model based on set pair situation with the majority inclusion relation proposed by professor Ziarko, the
variable precision set pair situation rough set model is proposed. Then the definition of positive region similarity is given. A
heuristic attribute reduction algorithm based on positive region similarity is presented. The time complexity of the algorithm
is analyzed. Simulation experiment shows the effectiveness of the proposed method for incomplete information system with
noise.