The Model Of ?-? Neighborhood Rough Sets And Its Applications

In today's information explosion society,every industry and every second has a huge amount of data generated and stored.Efficient analysis and mining of these data can not only bring great economic benefits,but also promote the further development of science and technology.Neighborhood rough sets,as an effective tool for processing numerical data,has been widely used.In this paper,we propose a new neighborhood rough sets model called ?-? neighborhood rough sets model.Furthermore,based on this model,we construct an attribute reduction algorithm and a neighborhood multi-granularity classification algorithm.The specific work is as follows:1.We propose a new neighborhood rough sets model called ?-? neighborhood rough sets model.Most of the existing neighborhood granularity models can not describe the neighborhood of mixed class samples well when they are used to describe the classification ability of attribute subsets.Our proposed model combines the advantages of both ?-neighborhood and ?-nearest neighbor rough sets model and has better performance to deal with data with unbalanced distribution.However,due to the ?-nearest neighbor rough sets model does not have monotonicity,so we define the rough approximation of decision for the proposed model by employing an iterative strategy,and we discuss the relevant properties.2.Based on the neighborhood rough set model,we construct an attribute reduction algorithm.Attribute reduction is an effective method for processing high-?imensional data.It can delete redundant features from data and keep the same classification ability with the system unchanged.The useful information in the data can be obtained by attribute reduction,which simplifies the process of knowledge processing.In this paper,we construct a forward greedy search algorithm,the experiment analysis shows that the proposed algorithm has better performance compared with some existing algorithms,especially than ?-neighborhood rough sets model and ?-nearest neighbor rough sets model.3.Based on the neighborhood rough sets model proposed in this paper,we also construct a neighborhood multi-granularity classifier.Classification problem is one of the most important research directions of data mining at present.It can classify unknown samples by mining known sample data.It is widely used in real life and production.The multi-granularity method is widely used in many fields such as data mining and pattern recognition.The multi-granularity method can mine the existing data from many angles,and can effectively improve the classification ability.Through experimental analysis,it is also proved that our proposed classification algorithm is more efficient than the existing classical classification algorithms and more precise.